Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Socioecon Rev 2015 Alvarez 449 75

Download as pdf or txt
Download as pdf or txt
You are on page 1of 27

Socio-Economic Review, 2015, Vol. 13, No.

3, 449475
doi: 10.1093/ser/mwv007
Advance Access Publication Date: 13 May 2015

Ignacio Alvarez*
Department of Applied Economics, University of Valladolid (Spain), Segovia, Spain/Complutense
Institute of Interational Studies (ICEI), Madrid, Spain
*Correspondence: nacho.alvarez@icei.ucm.es

Abstract
Using rm-level data for the period 2004 to 2013, this article examines the connection
between the nancialization of French corporations and functional income distribution in the non-nance sector of the economy. Financialization of French non-nancial
corporations has increased their dependence on earnings through nancial channels,
and diminished labor bargaining power in income distribution. We examine the
effects of these nancial revenues on wage share using a panel data model of 6980
French non-nancial rms. We conclude that increased dependence on nancial
prots is likely to decrease wage share in non-nancial corporations. Moreover, this
variable is more inuential in our model than the other variables usually identied by
the literature as determinants of functional income distribution, such as trade openness or labor market institutions. Of the determinants traditionally emphasized by the
literature, only technological change has a greater impact than nancialization.
Key words: nancialization, income distribution, wages, rms, France
JEL classication: E25 aggregate factor income distribution, E44 nancial markets and the
macroeconomy, J30 wages, compensation, and labor costs

1. Introduction
When analyzing the most important economic developments over the past few decades in
OECD countries two economic phenomena prove decisive: nancialization of the economy
and increased income inequality.
Both issues have generated considerable literature in recent years. However, with some
notable exceptions (Stockhammer, 2009, 2013; Onaran et al., 2011; Hein, 2012; Dnhaupt,
2013, 2014; Lin and Tomaskovic-Devey, 2013), there has been little previous study of the
connections between these two economic developments.
What are the links between nancialization and the evolution of income distribution? How
have wage dynamics been affected by nancialization trends? My research focuses on this

The Author 2015. Published by Oxford University Press and the Society for the Advancement of Socio-Economics.
All rights reserved. For Permissions, please email: journals.permissions@oup.com

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

Financialization, non-nancial corporations


and income inequality: the case of France

450

I. Alvarez

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

particular blind spot in the literature and aims to connect developments in the nancial
sphere of the economy with labor market and corporation functioning. Specically, focusing
on non-nancial corporations, I examine the connection between the nancialization of the
economy and the decrease in wage share for the French case.
I focus on France as one interesting case study of varieties of nancialization. Over the
past three decades, France has experienced one of the most intense liberalization and nancialization processes of any OECD country. This makes France an appropriate case to study the
consequences of nancialization on income distribution.
The signicant increase in inequality occurring in many OECD countries over the past
decades has spawned numerous studies on personal income distribution. Yet only recently
has mainstream economics begun to undertake substantial specic research addressing functional income distribution (European Commission, 2007; IMF, 2007a,b; OECD, 2007,
2008). This literature (see also Bentolila and Saint-Paul, 2003; Ellis and Smith, 2007) has
pointed to technological change, globalization and transformation in labor market institutions as key determinants of the wage share decline, with the rst two factors proving the
most relevant in understanding this trend.
Only recently and from the eld of post-Keynesian literature (Hein, 2011, 2012;
Dnhaupt, 2013, 2014; Stockhammer, 2013), has the phenomenon of nancialization been
considered a possible determinant of falling wage shares in OECD countries, since it modies
the bargaining power of the various actors.
The concept of nancialization has become widespread in political economy and refers in one
way or another to the growing inuence that nancial forces, markets and investors have had on
world economic dynamics since the late 1970s (Dumnil and Lvy, 2004, 2011; Stockhammer,
2004; Crotty, 2005; Epstein, 2005; Krippner, 2005; Palley, 2007; Hein, 2008, 2012; Orhangazi,
2008; Dallery, 2009, 2010; Lapavitsas, 2009; Bellamy Foster, 2010). Nonetheless, in the specialized literature there is no single common denition for this phenomenon.
Financialization can be examined from diverse perspectives that emphasize differing
aspects, and the various denitions need not be seen as mutually exclusive. Some authors
have highlighted the rapid liberalization of international nancial markets, the growing
instability of these markets, the de-intermediation and mercantilization of nancial systems
traditionally centered on banking, the formation of enormous stock market and credit
bubbles or the impact that all of this has on the macroeconomic functioning of national economies (Palley, 2007; Hein, 2008, 2012; Lapavitsas, 2009).
Other authors pay particular attention to the enormous weight institutional investors now
have in the global economy (Chesnais, 2004; Goyer and Hanck, 2006) and the parallel
process of international nancial liberalization. A further stream of inquiry emphasizes the
development of new corporate governance models based on maximizing shareholder value
and nancial protability in close association with the expansion of capital markets and
institutional investors (Aglietta, 2000; Plihon, 2004; Stockhammer, 2004; Dallery, 2009).
Whatever the case, nancialization can broadly be dened as two interdependent processes
(Lin and Tomaskovic-Devey, 2013): on one side, the growing ascendancy of the nancial
sector and its conception of control in the economy, and on the other the increasing participation of non-nancial corporations in capital markets and nancial services.
Most studies on nancialization, even those that have sought to explore the relationships
between the nancial and the non-nancial sectors, have focused on the former of the two processes mentioned. For example, Dumnil and Lvy (2004) focus their study on the income

Financialization, non-nancial corporations and income inequality

451

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

transferred from the non-nancial sector to the nancial sector for the main OECD economies.
Jayadev and Epstein (2007), Stockhammer (2004) and Dnhaupt (2012) analyze how the rentier
income share (in broad terms, interest, dividends and capital gains received by the owners of
nancial assets) has considerably increased in OECD economies during the past three decades.
In contrast to most studies on nancialization, and in line with the previous work by
Krippner (2005) or Lin and Tomaskovic-Devey (2013), I focus on the second process mentioned, that is, the increasing participation of non-nancial corporations in nancial
markets. I seek to shed some light on the process by which differences between nancial
and non-nancial corporations have blurred in recent decades, due to the emergence of
a pattern of accumulation in which prots accrue primarily through nancial channels
rather than through trade and commodity production (Krippner, 2005, p. 174).
Since my intention is to study the links between nancialization and functional income distribution, it is reasonable to focus the analysis on the meeting point of these two dimensions:
corporations. Large corporations are a particularly interesting sphere for analyzing the relations between nancial capital, prot generation and labor compensation, and therefore for
exploring the effect of nancialization on wage share.
Particularly, I analyze how the growing involvement of non-nancial companies in nancial markets, and the subsequent increase in nancial vis--vis non-nancial prots, has resulted in less bargaining power for employees relative to shareholders and managers. This
presumably results in a negative effect of nancialization on the wage share. I hope to contribute with this to a better understanding of the possible compromises or conicts in nonnancial rms between wage demands and the logics imposed by liberalized nancial markets.
Since the fundamental dimension I wish to address is the impact of the growing involvement of corporations in nancial markets, and the consequences of these corporations
increased nancial prots, I opted not to include nancial rms in the study.
My hypothesis, which is an extension to the French case of work carried out by Lin and
Tomaskovic-Devey (2013) for the US economy, states that increased dependence on earnings
through nancial channels is likely to decrease labors share of total income on non-nancial
corporations. Nevertheless, although said authors perform an industrial-level study, one contribution my work makes is to posit a rm-level approach.
I examine the effects of corporate nancial prots (as an exogenous proxy for nancialization) on wage share for the case of French non-nancial rms. For that purpose, I use a microeconometric approach based on the study of large French non-nancial corporations. This
approach consists of a panel data model of 6980 companies, from 2004 to 2013, and aims
to evaluate how the nancialization of corporate strategies affected salaries.
This article makes three main contributions to the literature. First, it is an extension to the
case of France of applied studies on nancialization, which tend to focus on Anglo-Saxon
economies (the USA and UK). Second, it is one of the few examples in the literature (together
with Dnhaupt, 2013; Lin and Tomaskovic-Devey, 2013; Stockhammer, 2013) to present an
econometric model to explain functional income distribution that includes a proxy variable
for nancialization. Third, the research adopts a relatively innovative methodology in studies
of both nancialization and analysis of functional income distribution; that is, it employs a
microeconometric approach using rm-level data. Certain studies have adopted this methodology to examine the consequences of nancialization on corporate investment (Orhangazi,
2008; Demir, 2009). I know of one only study that has used rm-level data to estimate the
determinants of wage share (Siegenthaler and Stucki, 2014). Yet such a methodology has

452

I. Alvarez

2. Some stylized facts: nancialization and income inequality


in the French economy
The many indicators of nancialization over the past 30 years include: (1) the enormous increase in the stock market capitalization of the main nancial centers, which now represents
several times the size of certain national gross domestic products (GDPs); (2) the fact that the
value of nancial transactions has increased much more rapidly than commercial or productive activity; (3) a signicant expansion in credit; and (4) the substantial increase in the relative
weight of nancial income vis--vis total national income.
Financial activity has expanded rapidly, while changes occurred in its relationship to production, its market composition, its products, the main agents involved and so on. In truth, we
cannot label nancialization as new, since it really involves certain features inherent to the
logic of capital that have intensied to the extreme in recent decades. Yet the expansive tendency
of nancial capital, exacerbated in this new context, has led it to rise above its strictly nancial
role and attempt to subject the whole realm of economic activity to its logic.

2.1 The nancialization of corporate strategies in France


Much of the literature exploring the nancialization of corporate strategies assumes nonnancial corporations to be passive agents that suffer the consequences of nancial liberalization and deregulation. Thus, numerous studies from a heterodox perspective (Dumnil and
Lvy, 2004; Stockhammer, 2004; Crotty, 2005; Hein, 2008, 2009; Orhangazi, 2008, Dallery,
2009, 2010) suggest that nancialization increases the payments made by corporations to nancial markets. These authors explain how, rather than being re-invested within the
company as gross xed capital formation, corporate prots are instead used for three different
types of payments: interest payments, dividend payouts and repurchase of company shares.1
This dynamic results from institutional changes in nancial markets, and the need for nonnancial corporations to adapt to the new context of nancial deregulation and free movement of capital. The extreme liquidity of international nancial markets offers immense
power to institutional investors, who can instantly sell stocks and thus punish corporations
that fail to meet the protability or management criteria established by these markets. Such
maneuvers can lead to sudden falls in a companys stock value, hostile takeovers and so on.
1

Many of these studies consider only gross ows and do not take into account net payments of
companies to nancial markets. An exception in this regard is the work of Skott and Ryoo (2007).

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

not previously been used to evaluate the effect of nancialization on functional income
distribution.
Following this introduction, the second section provides a review of the expansion of nancialization among the French non-nance sector, as well as the evolution of functional income
distribution in France, together with the most prevalent explanations. In the third section, I
discuss a possible theoretical framework to understand the links between the nancialization
of French corporate strategies and the evolution of functional income distribution. In the
fourth section, I discuss the specication, data, variables and econometric methods used to
estimate the model. The empirical results are presented and discussed in the fth section.
Finally, I summarize the main conclusions and their implications.

Financialization, non-nancial corporations and income inequality

453

2 Tobin pointed out that at a macroeconomic level investment in nancial assets cannot replace investment in real assets, since nancial assets ultimately relate to the productive investments of
companies.
3 It is reasonable to argue that an increase in nancial assets as a percentage of total assetsmainly
due to stock purchasesis over-valued as these stocks are accounted for at market prices. Firmin
(2008, p. 49) calculated the increase in stocks in corporate portfolios by volume at 1977 prices for
the French case. This calculation makes it possible to observe how the increase in nancial assets
over total non-nancial assets remains veriable, even though its value is logically inferior.

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

The new role that nancial investors have been playing since mid-1990s in the management of non-nancial corporations has led to profound changes in business strategies.
Anglo-Saxon institutional investors have acquired large equity stakes in French corporations,
achieving a spectacular degree of foreign penetration (Morin and Rigamonti, 2002; Plihon,
2004). They have also imposed new corporate governance criteria, which are now common
throughout France. This created a transition from a managerial or stakeholder type of management based on strong internal control of the company by the board of directors and managers to a shareholder model based on external control by liberalized nancial markets (Jeffers
and Plihon, 2001; Aglietta and Reberioux, 2004). Financialization has imposed a new corporate management model, based on maximizing shareholder value to comply with stronger
demands from nancial investors. The result has been higher payout ratios to capital markets.
However, large non-nancial corporations are more than merely passive actors subject to
the consequences of nancialization. They have also played an active role in nancializing
strategies and activities. Available company funds can be invested in acquiring either real
assets or nancial assets. When the business prospects in nancial markets are signicantly
better than those in goods and services markets, as has been the case in France over the
past two decades, there is a strong incentive for corporations to move investments from real
assets to nancial assets. Hence, nancialization has generated new prot sources for
non-nancial corporations, reinforcing the tendency to underinvestment of gross xed
capital formation (Stockhammer, 2004).
This potentially contradictory relationship between nancial and real investment was identied as far back as the 1960s in the work of Tobin (1965). Other studies (Stockhammer,
2004, 2007; Crotty, 2005; Krippner, 2005; Palley, 2007; Orhangazi, 2008; Milberg and
Winkler, 2010) indicate that increased investment in nancial assets by non-nancial corporations tend to crowd out real investment.2 According to Crotty (2005), this shift to the nancial
sphere was due to a stagnation of corporate prots in the productive sector throughout the
1980s and 1990s. Moreover, the adoption of new corporate governance criteria imposed a
corporate preference for short-term prots (Stockhammer, 2004). Financialization pushed
the directors of non-nancial corporations to behave in part like nancial investors.
When studying the USA, Krippner (2005) considered the increase in nancial investments
by non-nancial corporations in recent decades to be the main indicator of corporate nancialization. The same process has occurred in France, leading the amount of nancial assets in
the hands of French non-nancial corporations to increase sharply (Figure 1). In 1978, nancial assets accounted for only 36.4% of total assets, but by 2013 had reached 59% of the total.
What sort of activities were at the heart of this spectacular increase? Figure 2 indicates that
French rms increased their nancial investments in stocks, especially those of other corporations,3 while also offering loans to other non-nancial corporations.

454

I. Alvarez

Figure 2. Composition of NFC portfolios of nancial assets, 19782013 (total % of nancial assets).

As might be expected, a strong increase in nancial investments means that an increasing


percentage of company income is derived from nancial activities. Dividends received increase
due to the amount of shares held in other corporations, while greater amounts offered in loans
and credits generate more income from interest.
Figure 3 shows how the nancial income of non-nancial corporations as a percentage of
gross operating surplus increased from 5% to 8% during the 1950s and 1960s to 60% in
2013 (reaching 73.1% in 2008). This is highly illustrative of the concept of nancialization

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

Figure 1. Financial assets of non-nancial corporations (NFCs), 19782013 (% of total assets).

Financialization, non-nancial corporations and income inequality

455

mentioned in the introduction, understood as a pattern in which prots for corporations are
increasingly generated through nancial rather than productive activities.

2.2 The evolution of functional income distribution


Parallel to the nancialization of corporate strategies, a signicant increase in functional
income inequality has also emerged in France. Whereas in France the growth of inequality
in personal income distribution has been signicantly lower than in other OECD economies,
the pattern followed by functional distribution has been similar to that of other countries.
Figure 4 shows the evolution of functional income distribution in non-nancial corporations.
Therefore, France is also part of the trend of wage share decline displayed by other OECD
economies, especially in continental Europe. In the euro area, even if personal distribution has
remained comparably stable in recent years (as opposed to what has happened in Anglo-Saxon
economies), wage shares have decreased by around 10 percentage points of GDP
(Stockhammer, 2009). This decline is very similar to the evolution of the ratio between employee compensation to gross value added for French non-nancial corporations. Real wage
growth in French non-nancial corporations has clearly lagged behind productivity growth
since the 1980s, which has led to a profound change in the income distribution pattern
between capital and labor, which remained quite stable after World War II.
As noted by Stockhammer (2013), these changes in income distribution have led to a
renewed interest in analyzing the evolution of functional distribution, among not only heterodox but also mainstream economists. In general, and as mentioned in the introduction, this
literature (Bentolila and Saint-Paul, 2003; Ellis and Smith, 2007; European Commission,
2007; IMF, 2007a,b; OECD 2007, 2008; ILO, 2011) has identied technological change,
globalization and transformation in labor market institutions as key determinants to

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

Figure 3. Financial income of NFCs, 19502013 (% gross operating surplus).

456

I. Alvarez

explain the wage share decline. According to this literature, labor market institutions have
played a minor role, although this is not the case for the other two determinants.4
The basic orthodox argument considers that skill-biased technological change is the main
driver for changes in income distribution, since the rapid pace of progress in information technology has increased the demand for highly skilled and educated labor as opposed to lowskilled labor and has also become capital-augmenting rather than labor-augmenting. This
has shifted income distribution toward skilled labor and capital, leading total wage share
to decline.
External trade liberalization also has a signicant effect on understanding changes in the
functional distribution of income. According to this literature, since developed economies are
capital-abundant, this factor will gain from international trade. Besides, as been pointed out
by Stockhammer (2013), international trade benets the more mobile factor, usually capital,
undermining the bargaining position of labor in advanced countries.
The third factor usually associated with falling wage shares in the literature is labor market
institutions. Since wages and income distribution are largely the result of different collective
bargaining frameworks, in which different social groups claim their share of the added value,
the various institutions that shape these frameworks are key to understanding the evolution of
functional distribution. Thus, union density and coverage, employment protection legislation
or unemployment benets are variables that may affect income distribution, even if worker
bargaining power is a difcult dimension to capture.
Although mainstream literature reports strong effects of technological change and globalization on income distribution for OECD countries, in recent years there have been interesting
studies from a post-Keynesian perspective (Stockhammer, 2009, 2013; ILO, 2011) that have
4

For an interesting review of the determinants of income distribution in mainstream economics see
Stockhammer (2013).

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

Figure 4. Wage share in NFCs, 19502013 (compensation of employees/gross value added).

Financialization, non-nancial corporations and income inequality

457

called these ndings into question. These studies report that the effects of technological change
and globalization are not robust enough to explain the evolution of wage share, and they also
embrace other determinants of interest. Among these determinants one is of particular interest
to our research: nancialization which, according to these studies, proves key towards understanding changes in the bargaining power of workers. Stockhammer (2013) indicates that
nancialization has been the main cause of the decline in wage share, even more so than
technological change or globalization.

As pointed out in Secction 1, there is growing and interesting post-Keynesian literature exploring the links between nancialization and changes in functional distribution (Stockhammer,
2009, 2013; Onaran et al., 2011; Hein, 2012; Dnhaupt, 2013, 2014). This literature examines certain channels of inuence between nancialization and falling wage share. According
to these studies, the main link between nancialization and functional distribution occurs
through employee bargaining power.
Financialization of corporate strategies in non-nancial rms has led to a restructuring of
social relations between owners, managers and workers, reshaping the relative power of the
various actors. This phenomenon has had a dual effect on the bargaining position of labor
(Stockhammer, 2013). First, rms have gained not only geographical mobility but also mobility in terms of investment content, since companies can now invest more easily in nancial
assets (not only in real assets).
Second, nancialization has placed additional constraints on rms, due to greater pressure
on payout ratios from nancial markets (Stockhammer, 2004). These additional claims make
wages the variable that can be adjusted within the company to relieve tensions between the
nancial and the productive sphere. In corporations that experience less pressure from nancial markets, there is greater room to maneuver, a larger capacity for negotiation and possibilities of higher wages. Thus, nancialization becomes a lever for social recomposition that
alters relations between social classes.5
According to the rst of these two effects, the growing demand for maximizing shareholder
value in the short term has created the conditions for minimizing xed capital investment.
International capital market liberalization and expansion has imposed new corporate strategies on rms, leading to a transition from retaining and reinvesting to a new strategy
based on downsizing and distributing (Lazonick and OSullivan, 2000; Batsch, 2002). To
obtain higher short-term returns, companies have undertaken restructuring strategies based
on a sharp reduction of assets and secondary business lines, refocusing on those segments
of their supply chain that provide the greatest added value. Divestments, relocations and
domestic and international outsourcing have weakened the bargaining power of labor.
Thus, an increase in shareholder value orientation might contribute to weakening trade
union bargaining power. Moreover, due to the increasing prot opportunities stemming
from nancial deregulation, non-nancial corporations started to shift their investments

5 In this regard, the work of Martin et al. (2008) proves particularly interesting, as it analyzes how the
nancial liquidity of capital markets imposes discipline on the negotiating capacity of labor.

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

3. Financialization and functional income distribution: exploring


the linkages

458

I. Alvarez

6 For a detailed analysis on this issue, see Hein (2012).

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

from real to nancial assets. Rather than investing in physical capital to expand production
and employment, companies allocate a greater part of their resources to nancial markets.
This reliance on nancial income has profound consequences for corporations in terms of
investment, internal growth and income distribution (Fligstein and Shin, 2007; Lin and
Tomaskovic-Devey, 2013). First, rms that pursued shareholder value strategies (refocusing,
divestments or outsourcing, among others) were likely to reduce employment and therefore
employee bargaining power. Second, corporation over-reliance on nancial incomes means
that these revenues are reallocated away from production units and workers toward the companys nancial division. This leads to reduced labor bargaining power, since a growing share
of income is beyond the reach of workers and collective bargaining and can therefore barely be
negotiated. Finally, the external character that nancial income has for non-nancial
companiesrelatively independent from labordissociates prots and production and reinforces the negotiating power of shareholders and managers in collective bargaining (Lin and
Tomaskovic-Devey, 2013).
Although I decided to focus our study on this rst effect of nancialization on employee
bargaining power, I also address the second to better illustrate the links. The Kaleckian
distribution theory is useful to frame this second effect (Hein and van Treeck, 2010;
Hein, 2012). According to this theory, functional income distribution is determined by
means of marking up unit costs of companies in a context of monopolistic competition.
Logically, the mark-up must cover the rms costs, including labor salaries and gross
prots. The determinants of this mark-up vary (the degree of concentration within the industry, relevance of price competition versus differentiation, trade union bargaining power,
among others),6 and not all are of interest for this work. There is, however, one that proves
particularly important to understanding potential channels of the inuence of nancialization on wages: overhead costs.
From the perspective of individual rms, interest payments as well as dividends and
retained prots may be seen as overhead costs or obligations (Hein and van Treeck,
2010). A demand for increased interest and dividend payments might therefore induce
rms to raise prices. Nevertheless, in a context of intense liberalization and globalization
of international tradewhere prices can be sticky upwardadopting new nancial
norms entailing increased payout ratios may drive unit labor costs down (Stockhammer,
2006; Dallery and van Treeck, 2011; Hein, 2012). Indeed, strategies for maximizing
shareholder value have led to increased payments (as dividends and interest) made by
companies to shareholders and rentiers. Likewise these strategies have also led to increased
payments to senior managers, to linking their interests with shareholder value orientation
( primarily through stock options). This pressure from shareholders and rentiers to
ensure rms nancial protability and shareholder value has put downward pressure
on wages.
Furthermore, substituting real investment for nancial investment decouples prots
from xed capital formation and production, increasing non-reinvested prots. In addition,
higher nancial demands on rms have also meant higher payout ratios and lower retained
earnings available for investment and job creation. Thus, it plausible that as nancialization advanced across specic rms and industries, the relative power of labor declined,
reinforcing the claims of shareholders and managers (Lin and Tomaskovic-Devey, 2013).

Financialization, non-nancial corporations and income inequality

459

In fact, in Figure 5 we can see how, in general terms, those French industries with a higher
rate of nancialization (measured here as non-reinvested prots) for the period just
before the crisis are also those that have experienced lower wage growth. We then see
the inverse relationship behavior between non-reinvested prots and wage growth.
The correlation coefcient between these two variables is statistically signicant and
equal to 0.3.
These two approaches offer relevant contributions to understanding the nancialization of
corporate strategies and the impact the process has on the wage-labor nexus. To empirically
test these links, the following section analyzes data on French non-nancial corporations to
see how nancialization may impact functional income distribution.

4. Financialization and income inequality: a microeconometric study


with panel data
4.1 Baseline specication, variables and data
I carry out an econometric estimation in an attempt to answer the main hypothesis and evaluate to what extent wage share may be higher in less nancialized corporations. To conduct this

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

Figure 5. Rate of nancialization and wage growth in French industries, 20002008 (average rate and
annual growth rate).

460

I. Alvarez

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

estimation of the effects of nancialization on functional income distribution, I specify an


econometric model with rm-level panel data. I try to capture the relationship between the
concept of nancialization used in this worka pattern of accumulation in which prots
accrue increasingly through nancial channels (Krippner, 2005) and the evolution of functional income distribution.
The most widely used income inequality measures are not available at the rm level, which
is why the vast majority of studies conducted in the eld are performed at the country or industrial level. Although industrial-level research is likely to provide informative estimates for
rm-level processes, I consider the pioneering experience of developing econometric estimations for rm-level data. The downside of this proposal is that it entails using more ambiguous
proxy variables than are used at the industry or country level. The strengths are undeniable:
rm-level data models not only report more robust and detailed estimates but also allow for a
better understanding of the microeconomics of nancialization.
The conventional measure of wage share at industry or country level involves dividing total
compensation of employees to value-added. However, given the idiosyncrasies of data at rm
level, this measure is subject to major conicts. For example, it is an indicator which is highly
sensitive to the business cycle when considered at the microeconomic level. Furthermore, the
main problem of this measure is that in a database with a large number of observations, there
may be many companies reporting a cost of employees higher than the added value of a certain
period. In our sample, this happens for 10% of the observations, which can lead to a signicant bias in the estimates.
At the corporate level I therefore use a more stable indicator to study functional
income distribution, similar to the wage to value-added ratio. This alternative indicator
is the ratio of wage to total company assets. In fact, I divide most of our explanatory
variables by total assets as a way to address the bias resulting from the existence of highly
differing company sizes in the database. This is a common method used to correct for heteroskedasticity in microeconometric studies involving corporate-level data (Fazzari et al., 1988;
Orhangazi, 2008). Variables that are not divided by total assets are divided by another
indicator of rm size (such as sales) and are presented as ratios to avoid biases associated
with rm size.
Nonetheless, the baseline specications presented in this article have also been estimated
using the wage to (wage + EBITDA) ratio as a dependent variable. As pointed out, this indicator has a more obvious connection with income inequality than the wage to assets ratio.
However, the results of these estimations were not signicant for all independent variables,
with R 2 values close to 0. These results are obviously heavily biased by the fact that prots
are negative during this period for many companies. Once I remove all those observations
in the sample for which the EBITDA is negative, I obtain very similar results to those attained
when using the wage to assets ratio, as reported in Table A1 in the appendix. In any case, I
prefer to keep the wage to assets ratio as the dependent variable in our model, since I thus
avoid the bias caused by eliminating observations with negative prots.
Econometric time series models need to use stationary variables, that is, with no trend and
a similar degree of dispersion in their various observations. By dividing the variables by total
assets (or employment or sales) I solve the excessive dispersion problem in the sample.
However, to ensure the tendency is eliminated, I can take the rst difference of each variable.
Although Fisher-type tests exclude the presence of unit roots in the variables, I also use this
second specication to compare results.

Financialization, non-nancial corporations and income inequality

461

WAGE = f KRATIO, XPORTREV, EMPLOY, ADDVAL, SALES, EBIT, INTERST, FINPROF

The baseline specication for our model is dened in the following expression, where i
7

Amadeus is a database from Bureau van Dijk Electronic Publishing containing comprehensive
information on around 21 million companies across Europe.

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

I compiled integrated time-series cross-section data at rm level for 20042013 from the
Amadeus database.7 This sample constitutes an unbalanced panel, since the corporations included were not required to provide data for all the periods. Nonetheless, panel data make the
analysis more robust and signicant, with sufcient observations to allow for a disaggregated
analysis by sectors and size. I use the companys nancial prots as the proxy variable for
nancialization, and therefore as the central explanatory variable to test our hypothesis.
This variable is calculated as nancial revenues minus nancial expenses (thus including
interest, dividends and capital gains).
A series of variables representing the most prevalent explanations of functional income
inequality in the literature are incorporated as controls in the model. To account for capital
and skill-biased technology change, I control for capital-intensity ratio as a proxy for changes
in technology. This proxy is commonly used in the literature, since changes in gross xed capital
formation usually come with technological change (entrepreneurs use the latest technology accessible). I use the xed assets to total assets ratio, although the use of the capital-labor ratio is
prevalent in the literature when searching for a proxy of capital intensity. I conducted the estimates with both ratios and the results are almost identical. Nevertheless, I prefer to use the xed
assets to total assets ratio to better compare the size of the different coefcients.
Alternative measures of technological change, such as R&D expenses, the use of ICT
capital, or investment in computer hardware and software over non-residential xed assets
investment might prove more appropriate, although Amadeus does not provide sufcient statistical information on these variables. To account for the effects of globalization on wage
share, I also control for companies export revenue.
It is very difcult to capture employee bargaining power using rm-level data, since I lack information on union density, collective bargaining or coverage. One possibility is to use industrylevel variables. However, in an attempt to keep all the variables at the enterprise level, I use the
ratio of number of employees to total assets as a proxy for workers bargaining capacity. Labor
market institutions that strengthen the bargaining power of workers (union density, collective
agreements and job protection) are likely to prove more important when employee numbers
are higher and when past bargaining processes have consolidated these institutions.
To complement these control variables, I included dimensions traditionally considered to be
explanatory of corporate wage levels, such as the rms productivity (represented by the ratio of
added value to total assets), competitiveness (given by the ratio of sales to total assets) as well as
shareholder and manager pressure to maximize protability (proxied by the prot to total assets
ratio). I also consider the nancial constraints of the company as a possible determinant of wage
share. Companies that have to deal with high interest payments resulting from high debt ratios
will likely have less scope for wage growth.
Taking all this into account, I estimate a wage share equation that includes variables for
technological change (KRATIO), globalization (XPORTREV), bargaining power (EMPLOY),
capital productivity (ADDVAL), market competitiveness (SALES), protability (EBIT), nancial constraints (INTERST) and nancialization (FINPROF):

462

I. Alvarez

represents the individual corporations surveyed and t the years:


WAGE 1KRATIOi;t 2XPORTREVi;t 3EMPLOYi;t 4ADDVALi;t
5SALESi;t 6EBITi;t 7INTERSTi;t 8FINPROFi;t
I used the following variables from the Amadeus database to estimate the model expressed in
this equation. All of them except cashow are divided by the total assets of the company to
address biases associated with rm size:
corresponding to the cost of employees variable, which includes wage costs
paid by corporations as well as social payments or costs.
KRATIO
corresponding to the xed assets variable.
XPORTREV corresponding to export revenues.
EMPLOY
corresponding to employees.
corresponding to the added value generated by the company in a specic time
ADDVAL
period.
SALES
corresponding to the net volume of sales.
EBIT
corresponding to the Earnings Before Interest and Taxes.
INTERST
corresponding to nancial interest paid by the company.
FINPROF
corresponding to the nancial prots (or losses) of the company.
I use an additional variable to estimate a variation of our baseline specication: CASHSALE,
dened as the ratio of cashow to sales. As a starting point, I expect the following causal
relationships in our model:
KRATIOWAGE < 0; XPORTREVWAGE < 0; EMPLOYWAGE > 0; ADDVALWAGE > 0;
SALESWAGE ><0; EBITWAGE < 0; INTERSTWAGE < 0; FINPROFWAGE < 0
As already explained, I expect the variables KRATIO and XPORTREV to have a negative effect
on wage share, as has been widely highlighted by the literature. I expected the EMPLOY variable
to be positively correlated with wages, given that the number of workers is directly related to the
bargaining power of the labor force. Greater employee bargaining power will lead to higher
wages and, when labor demand is inelastic, to an increase in wage share (Stockhammer,
2013). Similarly, corporations with higher levels of efciency (ADDVAL) are expected to translate their greater efciency into better wages.
I do not, however, have any clear hypothesis for the sign of the coefcient associated with
corporate competitiveness (SALES), since it seems to be related to wages in contradictory
ways. Highly competitive corporations will obtain high returns due to their position as a
market leader and will in turn be able to pay employees higher wages. Nevertheless, efforts
to reach corporate competiveness may involve action such as wage restraint or even wage
cuts. When this occurs, an increase in the sales to assets ratio will not necessarily be linked
to higher wages but instead may result from greater wage restraint.
I expect a negative relationship between corporate protability (EBIT) and wage share, as
the stylized macroeconomic facts for the French economy suggest how advances in recent
decades in corporate protability have not come so much from new investments but rather
from wage restraint (Firmin, 2008; Husson, 2008). Furthermore, in general terms and
other variables being equal, I should expect salaries and protability to behave in an opposite

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

WAGE

463

Financialization, non-nancial corporations and income inequality


Table 1. Summary statistics of regression variables
Obs.

Mean

SD

Min.

Max.

WAGE
KRATIO
XPORTREV
EMPLOY
ADDVAL
SALES
EBIT
INTERST
FINPROF
CASHSALE
TOTASSETS

42 853
49 803
20 648
26 900
39 923
46 417
50 670
47 448
48 898
41 426
51 192

0.2891738
0.4309619
0.5563083
0.0077499
0.3770898
1.731378
0.1754439
0.0152961
0.0074672
6.994159
920961.1

1.32123
1.737641
1.159714
0.0383936
1.691848
4.33598
35.69223
0.099901
1.332235
15.5243
6 512 447

1.258601
8.78e09
1.408796
1.39e07
6.399324
0.4732974
7186
0.3739502
280.5424
99.693
0.001

262.8505
382.1718
32.16945
3.736833
329.8012
833.4645
6.064701
10.99357
34.74376
100
5.63e+08

Source: Amadeus. Authors own calculations. n = 6980, T = 10.

manner when dealing with functional income distribution. I also expect companies that have
to deal with high interest payments (INTERST) to exercise wage constraint. Finally, I expected
the central variable for testing the impact of nancialization on wage share, nancial prots
(FINPROF), to display a negative coefcient for the reasons mentioned.
The sample I took from Amadeus only includes companies classied as very large companies8 in the database. Since the scope of the analysis is the non-nance sector economy, I
excluded from the sample all corporations with SIC codes between 6000 and 6499.
Companies with a SIC code between 6500 and 6599 (real estate) remain in the sample, as
do those with code 6712 (holding ofces) except for nancial corporations, since most of
them are the parent company of other manufacturing and service companies. After removing
nancial rms from our database, I obtained a sample size of 6980 corporations with the descriptive statistics listed in Table 1.

4.2 Model estimation, methods and main results


The baseline specication is the result of a delicate equilibrium: maintaining a large sample
and including robust variables to explain wage share evolution at the rm level. The main
results for this specication are presented in Table 2.
Since I work with panel data, my rst decision is to clarify whether xed-effects estimation
is preferable to random effects. In line with the result of the Haussman test as well as existing
previous studies in the literature reviewed, it seems appropriate to opt for the xed-effects
(estimate 2 in Table 2) rather than the random-effects model (estimate 1 in Table 2).
Nevertheless, the sample shows both heteroskedasticity and autocorrelation, despite the
combination of longitudinal and cross-section data, which mitigates the risk of these estimation problems. Using Wooldridge and Breusch-Pagan/Cook-Weisberg tests to detect
8 Amadeus denes very large companies as those meeting at least one of the following four criteria:
their operating revenues are above 100 million , their total assets are valued at or above 200 million ,
they have at least 1000 employees or they are listed on the stock exchange. Excluded from this category are all companies with operating revenue per employee ratio or total assets per employee under
100.

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

Variable

464

Table 2. Results for baseline specication and variations (wage to assets ratio as dependent variable)

KRATIO
XPORTREV
EMPLOY
ADDVAL
SALES
EBIT
INTERST
FINPROF
_cons

1
RANDOM

2
FIXED

3
PCSE

4
DRISK

5
DRISK_EXOG

6
IV

0.0443***
0.0037
0.0045***
0.0009
0.2023***
0.0161
0.8796***
0.0029
0.0039***
0.0006
0.8451***
0.0052
0.8124***
0.0267
0.7810***
0.0099
0.0201***
0.0022

0.0217***
0.0056
0.0104***
0.0016
0.1327***
0.0148
0.6467***
0.0049
0.0248***
0.0011
0.6570***
0.0075
0.5824***
0.0267
0.5755***
0.0103
0.0452***
0.0028

0.0460***
0.004
0.0038***
0.0006
0.1875***
0.0277
0.8913***
0.007
0.0026***
0.0005
0.8554***
0.0125
0.8576***
0.0306
0.7866***
0.0175
0.0174***
0.0024

0.0217**
0.0081
0.0104***
0.0012
0.1327***
0.0335
0.6467***
0.0472
0.0248***
0.0025
0.6570***
0.0423
0.5824***
0.0461
0.5755***
0.038
0.0452***
0.0095

0.0378**
0.0119
0.0135***
0.0027
0.1122
0.0613
0.4378***
0.0341
0.0327***
0.0013

0.1508***
0.0136
0.0007
0.0033
0.1787***
0.0306
1.1750***
0.0257
0.0012
0.0025
2.1117***
0.0722
1.1219***
0.0631
0.8592***
0.0352
0.0246***
0.0059

0.3432***
0.0562
0.3955***
0.0268
0.0486***
0.0103

KRATIO

7
DIFF

XPORTREV
EMPLOY
ADDVAL

I. Alvarez

0.0196
0.0117
0.0204***
0.0024
0.1064***
0.0243
0.4414***
0.0315

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

0.0405***
0.002
0.4712***
0.0281
0.3471***
0.0531
0.3559***
0.0257
0.0000
0.0012

EBIT
INTERST
FINPROF
_cons

N
n
R2
F
P-value

12 690
2479
0.6825

12 690
2479
0.6961
2925.3
0.0000

12 690
2479
0.9365

12 690
2479
0.6964
1840.9
0.0000

12 691
2479
0.4656
11278.5
0.0000

12 339
2479

Standard errors below coefcients.


*P < 0.05, **P < 0.01, ***P < 0.001.
When using xed-effects the reported intercept is the average value of the xed effects.
(1) Random-effects GLS regression.
(2) Fixed-effects (within) regression.
(3) Prais-Winsten regression, heteroskedastic panels corrected standard errors (PCSE).
(4) Fixed-effects regression with Driscoll-Kraay standard errors.
(5) Fixed-effects regression with Driscoll-Kraay standard errors (dropping EBIT).
(6) Fixed-effects (within) Instrumental Variables regression, with EBIT = CASHSALE.
(7) First differences specication, xed-effects regression with Driscoll-Kraay.

8110
2002
0.5467
1328.0
0.0000

Financialization, non-nancial corporations and income inequality

SALES

Source: Amadeus. Authors own calculations.

465

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

466

I. Alvarez

5. Discussion of results and conclusions


Using a rm-data panel for the period 20042013, this article examines the connection
between the nancialization of French corporations and rising income inequality in the nonnancial sector of the economy. The analysis shows how increased dependence on earnings
through nancial channels is likely to decrease wage share in non-nancial corporations. The
main results obtained from the different estimations conrm my initial hypotheses: nancial

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

autocorrelation and heteroskedasticity in panel data, respectively, both phenomena are found
to be present in the sample. The multicollinearity test reveals the absence of a linear relationship between the model variables.
It was necessary to nd an estimation method that corrects the heteroskedasticity and autocorrelation errors. The most efcient way was to model the functional form of both the heteroskedasticity and autocorrelation, so that more efcient and precise estimates could be
obtained for the parameters (Cameron and Trivedi, 2009). For this purpose, I can use both
a panel corrected standard errors (PCSE) estimation (estimate 3 in Table 2), or Driscoll and
Kraay robust standard errors (estimate 4 in Table 2).
Hoechle (2007) states that PCSE approach may become inappropriate when the panels
cross-sectional dimension N gets large compared with the time dimension T, as is the case.
Driscoll and Kraays approach eliminates these possible deciencies. This is why I nally
prefer Driscoll and Kraay robust standard errors for panel regressions estimated by xed
effects. In this case, the error structure is assumed to be heteroskedastic, autocorrelated up
to a certain lag and possibly correlated between the groups ( panels). These standard errors
are robust to general forms of cross-sectional (spatial) and temporal dependence when the
time dimension becomes large.
It could be argued that not all the variables used in our model guarantee the exogeneity
condition. Protability (EBIT) in particular might prove endogenous since, ceteris paribus,
the reduction in labor cost will naturally lead to increased protably. The instrumental
variablecentered test for endogeneity suggests that EBIT is endogenous, implying the
model might therefore be biased if it is included.
Estimate 5 in Table 2 drops that variable to check whether the main results of the analysis
are preserved in the absence of this variable. One can see how the main results of the estimates
are maintained. In fact, results still hold and are robust in a nested model in which I successively drop other variables, such as ADDVAL, KRATIO, INTERST, XPORTREV and SALES,
and keep only EMPLOY and FINPROF.
A more appropriate method to deal with the endogeneity of protability is to use an instrumental variables estimation. Estimate 6 in Table 2 presents the results of this estimation, in
which I used the CASHSALE variable as an instrument of EBIT. CASHSALE is dened as
the cash ow to sales ratio and gives an idea of the companys ability to turn sales into
cash (a good instrument to proxy protability without being endogenous to wages).
Finally, estimate 7 in Table 2 is the rst-difference estimation. Even though I rejected the
unit root hypothesis for the variables, this estimate is presented for comparison purposes.
The xed-effects regression with Driscoll and Kraay standard errors is the method used to
estimate for the various economic sectors (Table 3) and the various sub-samples according
to company size (Table 4).

467

Financialization, non-nancial corporations and income inequality


Table 3. Estimation results by economic sector (xed-effects regression with Driscoll-Kraay
standard errors)

KRATIO
XPORTREV
EMPLOY

SALES
EBIT
INTERST
FINPROF
_cons

N
n
R2
F
P-value

2
Manufact.

3
Services

0.0217**
0.0081
0.0104***
0.0012
0.1327***
0.0335
0.6467***
0.0472
0.0248***
0.0025
0.6570***
0.0423
0.5824***
0.0461
0.5755***
0.038
0.0452***
0.0095

0.0064
0.0115
0.0004
0.0042
0.0659***
0.0127
0.3893***
0.0402
0.0688***
0.0028
0.4370***
0.035
0.3996***
0.0426
0.3354***
0.0256
0.0344***
0.0077

0.0155
0.0107
0.0106***
0.002
1.0448*
0.5406
0.7157***
0.0381
0.0191***
0.0019
0.7268***
0.036
0.6607***
0.0386
0.6489***
0.0327
0.0296**
0.0103

12 690
2479
0.6964
2868.05
0.0000

4998
902
0.6279
583.19
0.0000

7244
1468
0.7644
134100.00
0.0000

Standard errors below coefcients.


*P < 0.05, **P < 0.01, ***P < 0.001.
SIC codes included in MANUFACT group are from 2000 to 3999, and from 4000 to 8999 for SERVICES.
Source: Amadeus. Authors own calculations.

prot coefcients had signicant negative values, indicating an inverse correlation with the
wage to assets ratio.
These results also indicate that nancialization is one of the main determinants of wage share
at rm level, with a strong negative relationship for all the estimates presented. I obtain for
French corporations similar results to those previously presented by Stockhammer (2013)
and Dnhaupt (2013, 2014) for OECD economies, or Lin and Tomaskovic-Devey (2013) for
various US industries.
Therefore, akin to what several studies have shown for Anglo-Saxon economies, nancialization of the French economy is a profound change that reaches beyond nancial markets. It
is a change that involves a wide-ranging process of social recomposition and a substantial shift
in the bargaining power of the various social actors, strengthening owners and managers
111positions and resulting in the growing exclusion of labor from revenue generating process.

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

ADDVAL

1
Total

468

I. Alvarez

Table 4. Estimation results by company size (xed-effects regression with Driscoll-Kraay


standard errors)

KRATIO
XPORTREV

ADDVAL
SALES
EBIT
INTERST
FINPROF
_cons

N
n
R2
F
P-value

2
GROUP_2

3
GROUP_3

4
GROUP_4

0.0049
0.0081
0.002
0.0052
4.1889**
1.6174
0.2615***
0.056
0.0516***
0.0023
0.3224***
0.0503
0.2797***
0.0794
0.2445***
0.0515
0.0317***
0.0083

0.0208
0.0145
0.0038
0.0042
5.3000***
1.3087
0.4105***
0.0319
0.0451***
0.0026
0.4448***
0.0361
0.4322***
0.0426
0.3602***
0.0328
0.0303***
0.0073

0.0155
0.0126
0.0141**
0.0045
0.4094
0.3062
0.4692***
0.0553
0.0387***
0.0042
0.5070***
0.0408
0.4564***
0.0539
0.4674***
0.041
0.0759***
0.0163

0.0256
0.0145
0.0031**
0.001
2.1265**
0.6465
0.8087***
0.026
0.0069***
0.0012
0.8096***
0.0305
0.6863***
0.0562
0.6929***
0.0485
0.0201
0.011

2500
624
0.4766
777.21
0.0000

4359
1175
0.6034
5478.33
0.0000

3180
961
0.5694
6462.48
0.0000

2651
764
0.8624
3921.78
0.0000

Standard errors below coefcients.


*P < 0.05, **P < 0.01, ***P < 0.001.
Group 1: total assets above 300 million.
Group 2: total assets between 100 and 300 million.
Group 3: total assets between 50 and 100 million.
Group 4: total assets below 50 million.
Source: Amadeus. Authors own calculations.

As reported in Dnhaupt (2014) and Stockhammer (2013), the fundamental conclusions


reached by mainstream economists should be enhanced by taking into consideration new variables. Technological change (KRATIO), globalization (XPORTREV) and labor bargaining
power (EMPLOY) have the expected signs and are statistically signicant. However, they
cannot be said to be the only nor the main determinants of the evolution of income distribution.
To correctly illustrate the relative size of the coefcients in my estimates, I calculate the
marginal effects on the wage share of a 1 percentage point increase in the main explanatory
variables. Figure 6 presents the effects on wage share of this 1 percentage point increase in

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

EMPLOY

1
GROUP_1

Financialization, non-nancial corporations and income inequality

469

technological change (KRATIO), globalization (XPORTREV), labor bargaining power


(EMPLOY), interest paid by rms (INTERST) and nancialization (FINPROF). I focus on
the main explanatory variables present in the baseline estimation and omit the rest. Later I
address the impact of protability on wage share at rm level when correcting for possible
endogeneity.
To compute these marginal effects, I multiply the value of the coefcients by the median of
the respective underlying variable. The median is preferred here to the average, because the
latter is extremely sensitive to possible outliers. If calculations were carried out for the hypothetical average rm of the sample, results would surely be biased, since I have 6980 very different companies. For this reason, I calculate the marginal effects for the hypothetical median
rm in the sample.
Figure 6 shows that a one percentage point increase in the capital-intensity ratio (KRATIO)
would lead to a decrease of 0.8% point in the wage share. A 1%-point increase in the interest
paid by rms (INTERST), and in the nancialization ratio (FINPROF) would reduce wage
share by 0.5% and 0.4% point, respectively. An increase of 1 percentage point in the ratio
of trade openness (XPORTREV) would bring about a reduction of 0.2 percentage point in
the wage share. Finally, a similar increase in labor bargaining power (EMPLOY) would
lead to only a small increase in wage share (0.1 percentage point), even though this factor
tends to be widely stressed by both orthodox and heterodox analysis. All the estimates presented more or less share these results, including regression with instrumental variables and
estimation in rst differences.
These results show that among the determinants traditionally emphasized by the literature
to explain the decrease in wage sharetechnological change, trade openness and labor market
institutionsonly the rst has a greater effect on income distribution when compared with
nancialization. The results obtained in my estimates indicate that to explain the differences

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

Figure 6. Marginal effects on the wage share of a one percentage point increase in main explanatory
variables (%).

470

I. Alvarez

9 It is important to note that remuneration by stock options is only offered to a minority of non-executive
employees, but may constitute a sufcient amount to alter the estimates.

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

in wage shares at the rm level one must take into account other dimensions that are not
usually considered, particularly those related to the nancial sphere.
Besides the nancialization effect already mentioned, and the impact of interest paid by
rms, coefcients for protability (EBIT) are also signicant, of relevant magnitude and opposite in sign to wage share. When analyzed in the instrumental variable regression specication, corporate protability strengthens its signicant negative effect on wages. In fact, when
we calculate the marginal effect of protability (EBIT) on wage share we see how a one percentage point increase leads to a decrease of 2.2% points in wage share. It must be understood
that unlike what happens when the two components of the functional distribution at the
macroeconomic level are analyzed, in our model both wages and prots might grow simultaneously, since they are divided by total company assets, not by added value. This possibility
shows how the strong negative relationship between the two variables pinpointed responds
not only to possible endogeneity.
When I break down my estimates by economic sectors (Table 3), I nd that the main
conclusions of my analysis are still supported. The coefcients associated with the proxy
variable representing nancialization show negative values for both the manufacturing
sector and for services, with greater intensity in the latter. This difference probably accounts
for a more gradual advance of the nancialization process among industrial companies compared with service companies. All the other variables analyzed have similar coefcients compared with the estimates for the database as a whole, both for manufacturing and services.
When the sample is divided into different company sizes, the sign of the coefcients
remains unchanged. Table 4 indicates that the coefcients evaluating the impact of nancialization on wage share are clearly signicant and negative for the four groups of corporations
analyzed. However, I also observe that the values of these coefcients are somewhat smaller in
the largest corporations (groups 1 and 2). This less signicant repercussion may be due to two
phenomena. First, larger corporations constitute more solid arenas for workers to enter wage
negotiations, which makes it more difcult to lose bargaining power, and therefore to transfer
wages to protability claims. In fact, in the largest corporations (groups 1 and 2) one can see
how lower coefcients for the nancialization proxy are accompanied by signicantly higher
values for bargaining power of workers and less impact of corporate protability on wage
share compared with smaller corporations (groups 3 and 4). In contrast, the smallest rms
in the sample experience a higher impact of nancialization, lower bargaining power of
workers and greater inuence of protability demands.
The second phenomenon has to do with the fact that corporations in groups 1 and 2 actively develop plans to remunerate part of their non-executive employees by means of stock
options and similar instruments, which are included in corporations labor costs. The statistical result of this would be a less regressive impact of nancialization on wage share.9
The results obtained in this research remain unchanged, as already mentioned, when I use
the capital-labor ratio as a proxy for technological change, instead of using xed assets to total
assets ratio. Moreover, the results also hold when I use the wage to (wage + EBITDA) ratio as a
dependent variable, as seen in Table A1 in the appendix.
Since at rm level functional income distribution is not only determined by technological
progress, the weight of globalization or labor market institutions, one must consider

Financialization, non-nancial corporations and income inequality

471

References
Aglietta, M. (2000) Shareholder Value and Corporate Governance. Some Tricky Questions,
Economy and Society, 29, 146159.
Aglietta, M. and Reberioux, A. (2004) Drives du Capitalisme Financier, Paris, Albin Michel.
Batsch, L. (2002) Le Capitalisme Financier, Paris, La Dcouverte.
Bellamy Foster, J. (2010) The Financialization of Accumulation, Monthly Review, 62, 117.
Bentolila, S. and Saint-Paul, G. (2003) Explaining Movements in the Labor Share, B.E. Journal of
Macroeconomics, 3, 133.
Cameron, C. A. and Trivedi, P. K. (2009) Microeconometrics Using Stata, Texas, Stata Press.
Chesnais, F. (ed.) (2004) La Finance Mondialise. Racines Sociales et Politiques, Conguration,
Consquences, Paris, La Dcouverte.
Crotty, J. (2005) The Neoliberal Paradox. The Impact of Destructive Product Market Competition
and Modern Financial Markets on Nonnancial Corporation Performance in the Neoliberal
Era. In Epstein, G. (ed.) Financialization and the World Economy, London, Edward Elgar,
pp. 77110.
Dallery, T. (2009) Post-Keynesian Theories of the Firm under Financialization, Review of Radical
Political Economics, 41, 492515.
Dallery, T. (2010) Le divorce rentabilit/croissance dans le capitalisme nanciaris. Changements
de rgimes, quilibres, instabilits et conits, PhD Thesis, Universit Lille 1.
Dallery, T. and van Treeck, T. (2011) Conicting Claims and Equilibrium Adjustment Processes in
a Stock-ow Consistent Macroeconomic Model, Review of Political Economy, 23, 189211.
Demir, F. (2009) Financial Liberalization, Private Investment and Portfolio Choice.
Financialization of Real Sectors in Emerging Markets, Journal of Development Economics,
88, 314324.
Dumnil, G. and Lvy, D. (2004) Capital Resurgent. Roots of the Neoliberal Revolution,
Cambridge, MA, Harvard University Press.
Dumnil, G. and Lvy, D. (2011) The Crisis of Neoliberalism, Cambridge, MA, Harvard University
Press.
Dnhaupt, P. (2012) Financialization and the Rentier Income ShareEvidence from the USA and
Germany, International Review of Applied Economics, 26, 465487.
Dnhaupt, P. (2013) The Effect of Financialization on Labors Share of Income, IPE Working
Papers 17/2013, Berlin School of Economics and Law, Institute for International Political
Economy (IPE).

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

additional factors to addresses the problem of inequality. The distribution struggle taking
place within non-nancial companies is also closely related to the structure and incentives
of the nancial system. These are the factors to be addressed from the economic policy standpoint if we wish to reverse the downward trend of the functional income distribution.
The relevance and specic nature of the measure of nancialization presented in this article
requires further investigation. My research suggests that prots in non-nancial corporations
increasingly accrue through nancial channels, which results in eroded labor bargaining
power and lower wage share. Nevertheless, future research should consider to what extent
this measure of nancialization goes beyond the fact that non-nancial corporations imitate
nancial institutions making revenues out of nancial channels. Studying to what extent nancial revenues are masking a new mode of organizing production chains for non-nancial
corporations, through ownership of subsidiary companies that are not consolidated, would
allow a better understanding of the erosion in labor bargaining power.

472

I. Alvarez

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

Dnhaupt, P. (2014) An Empirical Assessment of the Contribution of Financialization and


Corporate Governance to the Rise in Income Inequality, IPE Working Papers 41/2014, Berlin
School of Economics and Law, Institute for International Political Economy (IPE).
Ellis, L. and Smith, K. (2007) The Global Upward Trend in the Prot Share, Working Papers No.
231, Bank for International Settlements (BIS).
Epstein, G. (ed.) (2005) Financialization and the World Economy, London, Edward Elgar.
European Commission. (2007) The Labour Income Share in the European Union, in , Employment
in Europe 2007, Brussels, European Commission.
Fazzari, S., Hubbard, R. and Peterson, B. (1988) Financing Constraints and Corporate Investment,
Brookings Papers on Economic Activity, 1, 141195.
Firmin, C. (2008) Financiarisation, rpartition des revenus et croissance en France. Quelques faits
styliss lpreuve dun modle stock-ux, PhD Thesis, Universit Paris I Panthon-sorbonne.
Fligstein, N. and Shin, T. (2007) Shareholder Value and the Transformation of the US.Economy,
19842000, Sociological Forum, 22, 399424.
Goyer, M. and Hanck, B. (2006) Labour in French Corporate Governance. The Missing Link, in
Gospel, H. and Pendleton, A. (eds) Corporate Governance and Labour Management. An
International Comparison, Oxford, Oxford University, pp. 173196.
Hein, E. (2008) Financialisation in a Comparative Static, Stock-ow Consistent Post-Kaleckian
Distribution and Growth Model, IMK Working Paper 212008, Hans Boeckler Foundation,
Macroeconomic Policy Institute.
Hein, E. (2011) Distribution, Financialisation and the Financial and Economic Crisis
Implications for Post-crisis Economic Policies, MPRA Paper 31180, University Library of
Munich.
Hein, E. (2012) The Macroeconomics of Finance-Dominated CapitalismAnd its Crisis, London,
Edward Elgar.
Hein, E. and van Treeck, T. (2010) Financialisation and Rising Shareholder Power in Kaleckian/
Post-Kaleckian Models of Distribution and Growth, Review of Political Economy, 22, 205233.
Hoechle, D. (2007) Robust Standard Errors for Panel Regressions with Cross-sectional
Dependence, Stata Journal, 7, 281312.
Husson, M. (2008) Un pur Capitalisme. Lausanne, Page Deux.
ILO. (2011) The Labour Share of Income: Determinants and Potential Contribution to Exiting the
Financial Crisis, in World of Work Report 2011: Making Markets Work for Jobs, Geneva,
International Labor Organization.
IMF. (2007a) The Globalization of Labor, in World Economic Outlook 2007, April, Washington,
DC International Monetary Fund.
IMF. (2007b) Globalization and Inequality, in World Economic Outlook 2007, October,
Washington, DC, International Monetary Fund.
Jayadev, A. and Epstein, G. (2007) The Correlates of Rentier Returns in OECD Countries, Working
Papers Series No. 123, Political Economy Research Institute, University of Massachusetts at
Amherst.
Jeffers, E. and Plihon, D. (2001) Investisseurs institutionnels et gouvernance des entreprises, Revue
dEconomie Financire, 63, 137152.
Krippner, G. (2005) The Financialization of the American Economy, Socio-Economic Review,
3, 173208.
Lapavitsas, C. (2009) Financialisation, or the Search for Prots in the Sphere of Circulation,
Discussion Papers Series No. 10, Research on Money and Finance, Department of Economics,
SOAS.
Lazonick, W. and Sullivan, M. (2000) Maximizing Shareholder Value. A New Ideology for
Corporate Governance, Economy and Society, 29, 1335.

Financialization, non-nancial corporations and income inequality

473

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

Lin, K. H. and Tomaskovic-Devey, D. (2013) Financialization and U.S. Income Inequality,


19702008, American Journal of Sociology, 118, 12841329.
Martin, R., Rafferty, M. and Bryan, D. (2008) Financialization, Risk and Labour, Competition
and Change, 12, 120132.
Milberg, W. and Winkler, D. (2010) Financialisation and the Dynamics of Offshoring in the USA,
Cambridge Journal of Economics, 34, 275293.
Morin, F. and Rigamonti, E. (2002) volution et structure de lactionnariat en France, Revue
Franaise de Gestion, 141, 155181.
OECD. (2007) OECD Workers in the Global Economy: Increasingly Vulnerable?, in OECD
Employment Outlook 2007, Paris, OECD.
OECD. (2008) Growing Unequal? Income Distribution and Poverty in OECD Countries, Paris,
OECD.
Onaran, ., Stockhammer, E. and Gra, L. (2011) Financialization, Income Distribution, and
Aggregate Demand in the US, Cambridge Journal of Economics, 35, 637666.
Orhangazi, O. (2008) Financialization and Capital Accumulation in the Nonnancial Corporate
Sector. A Theoretical and Empirical Investigation on the US Economy, 19732003,
Cambridge Journal of Economics, 32, 863886.
Palley, T. (2007) Financialization. What it Is and Why it Matters, Working Paper Series No. 153,
Political Economy Research Institute, University of Massachusetts at Amherst.
Plihon, D. (2004) Le Nouveau Capitalisme, Paris, La Dcouverte.
Siegenthaler, M. and Stucki, T. (2014) Dividing the Pie: the Determinants of Labors Share of
Income on the Firm Level, KOF Working Papers No. 352, KOF Swiss Economic Institute.
Skott, P. and Ryoo, S. (2007) Macroeconomic Implications of Financialization, Working Paper
Series, 200708, University of Massachusetts at Amherst.
Stockhammer, E. (2004) Financialization and the Slowdown of Accumulation, Cambridge Journal
of Economics, 28, 719741.
Stockhammer, E. (2006) Shareholder Value Orientation and the Investment-Prot Puzzle, Journal
of Post Keynesian Economics, 28, 193215.
Stockhammer, E. (2007) Some Stylized Facts on the Finance-Dominated Accumulation Regime,
Political Economy Research Institute, Working Papers Series No. 142, University of
Massachusetts at Amherst.
Stockhammer, E. (2009) The Finance-Dominated Growth Regime, Distribution, and the Present
Crisis, Department of Economics Working Papers No. 127, Vienna University of Economics.
Stockhammer, E. (2013) Why Have Wage Shares Fallen? A Panel Analysis of the Determinants of
Functional Income Distribution, Conditions of Work and Employment Series No. 35,
International Labour Organization.
Tobin, J. (1965) Money and Economic Growth, Econometrica, 33, 671684.

474

Appendix
Table A1. Results for baseline specication and variations (wage to wage + EBITDA ratio as dependent variable)

KRATIO
XPORTREV
EMPLOY
ADDVAL
SALES
EBIT
INTERST
FINPROF
_cons

1
RANDOM

2
FIXED

3
PCSE

4
DRISK

5
DRISK_EXOG

6
IV

0.0420***
0.0037
0.0032***
0.0009
1.7995***
0.0825
0.8612***
0.0035
0.0023***
0.0006
0.7903***
0.0089
0.7693***
0.0271
0.6968***
0.0131
0.0147***
0.0022

0.011
0.0058
0.0088***
0.0016
1.1292***
0.0879
0.6300***
0.0056
0.0196***
0.0011
0.5618***
0.0109
0.5770***
0.0293
0.4992***
0.0138
0.0430***
0.0029

0.0498***
0.0044
0.0027***
0.0006
1.6237***
0.3034
0.8835***
0.0121
0.0011**
0.0004
0.8388***
0.0156
0.7936***
0.0387
0.7453***
0.0245
0.0153***
0.0025

0.011
0.0073
0.0088***
0.002
1.1292
0.6082
0.6300***
0.0484
0.0196***
0.003
0.5618***
0.0393
0.5770***
0.0535
0.4992***
0.0346
0.0430**
0.0141

0.0232**
0.0084
0.0092*
0.0037
1.5341
0.8719
0.5018***
0.0366
0.0177***
0.0027

0.3320***
0.0546
0.0015
0.0093
3.3764***
0.7565
2.0597***
0.1857
0.0236***
0.0064
6.0300***
0.7294
1.3641***
0.1967
1.4497***
0.1532
0.0967***
0.0188

0.5098***
0.0733
0.4010***
0.029
0.0331*
0.0145

KRATIO

7
DIFF

XPORTREV
EMPLOY

I. Alvarez

0.0126
0.0151
0.0233***
0.0034
0.4043*

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

SALES
EBIT
INTERST
FINPROF
_cons

N
n
R2
F
P-value

10 042
2243
0.6762

10 042
2243
0.6902
2169.53
0.0000

10 042
2243
0.9588

10 042
2243
0.6902
5274.21
0.0000

10 043
2243
0.5855
3720.30
0.0000

9918
2227

5912
1685
0.5146
1570.53
0.0000

Standard errors below coefcients.


*P < 0.05, **P < 0.01, ***P < 0.001.
When using xed-effects the reported intercept is the average value of the xed effects. All those observations in the sample for which the EBITDA is negative are removed.
(1) Random-effects GLS regression.
(2) Fixed-effects (within) regression.
(3) Prais-Winsten regression, heteroskedastic panels corrected standard errors (PCSE).
(4) Fixed-effects regression with Driscoll-Kraay standard errors.
(5) Fixed-effects regression with Driscoll-Kraay standard errors (dropping EBIT).
(6) Fixed-effects (within) instrumental variables regression, with EBIT = CASHSALE.
(7) First differences specication, xed-effects regression with Driscoll-Kraay.

475

Source: Amadeus. Authors own calculations.

Financialization, non-nancial corporations and income inequality

0.1913
0.4082***
0.0366
0.0354***
0.0033
0.3506***
0.0391
0.2776***
0.0512
0.2813***
0.0406
0.0002
0.0011

ADDVAL

Downloaded from http://ser.oxfordjournals.org/ at Universidade Federal da Bahia on September 8, 2016

You might also like