1A large consensus has arisen that, since the 1980s, the economy has been changed by the process of financialization, and that this process affects not only the modes of financial exchange, but also, more broadly, social and economic structure and inequalities (Fligstein 2001; Krippner 2005). The impact of financialization on social inequalities and cohesion is often handled in the media in spectacular terms such as “stock market layoffs,” comparing the impoverishment of some (the laid-off workers) with the enrichment of others (the capital gains of stockholders). While there can hardly be any doubt as to the existence of the layoffs, their causal relationship to the stock market is actually hazy – it is not clear that such layoffs would not have taken place in a less financialized economic system – and their impact on capital gains is uncertain (Capelle-Blancard and Couderc 2006). While financialization admittedly has repercussions on management of ordinary staff in nonfinancial companies (Montagne and Sauviat 2001), particularly regarding the sustainability of jobs and the use of layoffs or variable compensation, the causality involved in its impact through those channels on the structure of inequality is long, complex, and therefore uncertain.
2On the other hand, financialization potentially has a more direct impact on inequalities by reason of the categories of employees that it promotes: those in charge of financial matters, whether they are employed in the financial sector or work as finance specialists in nonfinancial companies. A number of studies have begun to show the size of incomes collected by employees in the financial sector and its role in increasing inequality (Philippon and Reshef 2009; Bell and Van Reenen 2010; Godechot 2012), especially in the United States, the United Kingdom, and France, but thus far they have barely begun to address the question of financialization outside the financial sector.
3Studies of inequality concentrating on national distributions of wage and income (Atkinson, Piketty, and Saez 2010; Piketty 2001) sometimes tend to give the impression that the increase of income and (especially) wage inequalities is a phenomenon that affects the entire economy, that is, all economic sectors and all regions indiscriminately. However, the financialization process does not apply equally throughout a given country, owing to the historic concentration of financial activity in major financial centers such as New York, London, and Paris (Sassen 2001, 2005). Nevertheless, the dematerialization and computerization of financial activity could potentially allow financial activity to be liberated from these historic centers, promoting a desegregation process. Saskia Sassen (2001, 2005) affirms that globalization and computerization are promoting the growth, opulence, and segregation of the major financial centers, rather than making a dent in them. They contribute to the emergence of “global cities” due to the added value offered by idiosyncratic information transmitted face to face rather than standardized information transmitted electronically, as well as the added value of strategic control and organization jobs over the classic delocalized production and distribution jobs. Following these hypotheses, this globalization/financialization movement would thus favor the growth of sociospatial inequalities, both between the global cities and the rest of the country as well as within the global cities themselves.
4This last hypothesis leads us to infer from this emergence of global cities an increase in spatial segregation, not only at work, but also in residential areas by income level. In the case of the United States, Sean F. Reardon and Kendra Bischoff (2011) have thus demonstrated a causal relationship between the increase in income inequality and spatial segregation, but they do not specify the areas most affected by this increased move toward polarization. One might suppose that the major financial centers, such as New York or London, play a role in this process, both because they have a concentration of financial jobs and because the high wages that they offer promote residential strategies of social avoidance (Maurin 2004).
5We propose to connect several areas of research on present-day inequality that remain too compartmentalized – income inequality (Atkinson, Piketty, and Saez 2010), financialization (Krippner 2005), social geography (Sassen 2001), and social segregation (Reardon and Bischoff 2011) – testing the positive relationships between financialization, the increase in wage inequality, the emergence of global cities, and social segregation at home and at work. We will examine these ties using administrative wage data from the DADS (Déclarations Annuelles de Données Sociales: “Annual Declarations of Social Data”) filed on employees by companies in the private sector, with representative data (1/24th of files) since 1976  and exhaustive data since 1994.
6The first part of this article will discuss indicators of financialization. The second will show that financialization and the increase in inequality primarily affect the Île-de-France region, while the third will show that such a movement promotes an increase in segregation at work and (to a lesser degree) in residential areas, but in all cases promotes increased social separatism among the most highly compensated employees. Finally, in the last section, we will try to clarify the contribution of financialization to this dynamic of social separatism.
1 – Financialization and Wage Structure
7Financialization is a complex and diverse process that has undeniably affected Western economies over the past thirty years, to the point that it has become the object of broad social protests in recent years (with the “Occupy Wall Street” movement), as well as an electoral campaign issue. The exact impact of this phenomenon and its full implications remain difficult to define, as it is manifested at so many different levels: from the impact of stock prices on politics to mortgage approval procedures, as well as accounting methods used to calculate the performance of firms.
8The socioeconomic literature on financialization (Deutschmann 2011; Tomaskovic-Devey and Lin 2011; Krippner 2005; Epstein 2005) distinguishes among several levels of financialization, depending on the sector (financial or nonfinancial) and the scale (at the macroeconomic level, the firm organization level, and the levels of the individual consumer and investor).
9Financialization is manifested first and foremost through an internal change in the modes of financial intermediation. While financial markets occupied a relatively anecdotal place in economic life in the 1960s, to the point that some commentators predicted their demise, institutional changes of the past twenty-five years have helped to place them once again at the heart of the action. First of all, the end of the fixed-exchange-rate regime in the early 1970s, based on the Bretton Woods Agreement, revived the exchange market. Faced with the inflation crisis of the 1970s (Krippner 2011), Western countries, particularly the United States, the United Kingdom, and France, introduced policies to revive the market. They suppressed and dismantled barriers deemed to be anticompetitive, such as fixed brokerage fees, the Compagnie des Agents de Change, or even the Glass-Steagall Act; they added many new products (especially derivatives) and market segments; and they participated in the computerization of transactions and promoted individuals’ participation in the stock market by favoring pension funds (Montagne 2006). The combination of this promarket policy and the game playing in the international exchange that promoted the accumulation of liquid assets (petrodollars, Chinese trade surplus, the highly accommodating monetary policy in the OECD countries starting in 2001) led people to turn more and more to financial markets. Transaction volumes soared spectacularly during the periods of 1977-1987, when the Paris stock exchange grew elevenfold (Godechot 2012), 1995-2000 (grew sevenfold), and 2004-2007 (doubled). This unprecedented expansion was accompanied by changes in financial behaviors, among which we will highlight growing support for risk quantification (McKenzie 2006) and for liquidity (Orléan 1999) – the possibility of reversing a financial position quickly and at minimal cost – short-termism, and the development of a generalized arbitration (with a speculative component) of related financial products.
10Financialization is not limited to finance alone; it leaves its mark on other economic sectors. First, the strategies and work organization specific to financial markets are being applied to a growing number of nonfinancial product markets called “commodities” (energy products, raw materials, agricultural products). Next, large firms are often equipped with internal trading floors to manage their cash flow and their risks, and to this end they employ small “trading” teams. They often dispose of banking subsidiaries in order to promote credit sales. More generally, beyond these niche financial professions, financialization is also changing management standards.
11The new standards are often spread by consulting firms. Some of them, like Stern Stewart & Co. with its patented accounting concept known as EVA (Economic Value Added), have played a widely recognized role in the dissemination of financial logic (Lordon 2000; Froud and Williams 2000). The model of the multidivisional firm, whose favored objectives were diversification and expansion, gave way to the model of the firm dedicated to creating shareholder value (Fligstein 2001), restricted to its core business, the goal of which is to maximize the company’s stock market value. The financial strategies of liquidity, risk measurement, benchmarking, arbitrage, and discounted cash flow were introduced into nonfinancial firms for the choice of investments, accounting valuation, or work organization, at the cost of classic industrial and business strategies. The company is broken down into as many profit centers as basic units, the value of each being measured in the light of its potential resale value on the market. A symptom of this sea change is the growing power of financial directors at the head of firms and their designation as “chief financial officer,” replacing chief operating officers (Zorn 2004; Zorn et al. 2005). Though this movement was initiated by a change in U.S. tax code (Zorn 2004), its continuation was promoted by the shareholder value ideology, the hostile IPOs of the 1980s, and the increased role of financial analysts in establishing stock prices (Zorn et al. 2005).
12At the macroeconomic level, Greta Krippner (2005) and, more recently, Donald Tomaskovic-Devey and Ken-Hou Lin (2011) have found the distribution of profits in the U.S. economy a better marker of financialization than the evolution of financial work could possibly be. The weight of the finance/insurance/real estate sector peaked at 45% of profits, whereas the same sector oscillated between 10% and 20% between 1950 and 1980 (Krippner 2005; Tomaskovic-Devey and Lin 2011). Krippner has also analyzed the growing portion of financial revenues in nonfinancial firms as a symptom of the financialization of the nonfinancial sector (2005). Tomaskovic-Devey and Lin (2011) have also analyzed the widening gap between average wages paid in the financial sector and those paid in the nonfinancial sector; they estimate that, between 1981 and 2008, six trillion dollars were transferred from the nonfinancial sector to the financial sector.
The Ascendancy of the Financial Sector
13Based on national accounting data, duplicating the financialization approach based on the distribution of profits does not produce similar results in the case of France. The proportion of total gross operating income, distributed income, and gross disposable income held by financial companies appears, on initial examination, stable.  Similarly, when one attempts to measure the ascendancy of the financial sector through the evolution of its numbers in the workforce, it is clear, as Krippner (2005) found, that this indicator is hardly a reflection of the massive financialization of the labor force. The proportion of the workforce employed in the financial sector has changed very little (Figure 1).  One of the reasons for this quantitative indicator’s lack of sensitivity is related to the heterogeneity of the financial sector. This sector includes traditional banks, retail banks, and banks that finance small firms, all of which have little to do with financial markets, and which could not be said to have been the vector of financialization; it also includes “corporate and investment banks,” which are particularly emblematic of that process. And yet, while corporate banks were adding employees, retail banks during the same period adopted a strategy of rationalization and staff reduction (Dressen and Roux-Rossi 1997). At the aggregate level, these two contradictory trends – difficult to separate, due to insufficiently detailed sector designations and the French universal-bank model, which includes retail, corporate, and investment banking under one label – cancel each other out.
14Tomaskovic-Devey and Lin (2011) nevertheless supplement the profit-based approach with an approach that considers wages, showing that wages in the financial sector have taken off by comparison with those in other sectors. However, as shown by Philippon and Reshef (2009), excessive wages in the financial sector are a direct consequence of financial deregulation measures in the United States. Approaching financialization not only through average wage levels, but also at the highest ends of the spectrum (Godechot 2012), could be the way to see the new centrality of finance, owing to its wage share. This leads us to use wage data, from the DADS (see box), to isolate the share and the effect of the employees who participate the most and, potentially, profit the most from the phenomenon of financialization. Working with individual data is particularly recommended when one wishes to show the impact of this phenomenon on inequality and the spatial distribution thereof. 
DADS (Déclarations annuelles de données sociales: “Annual Declarations of Social Data”): A Key Source for the Study of Wages in France
Because the numbers of hours declared are not always accurate, and because hourly wage is not necessarily the best approach for finding high wages (which in certain professions, such as consulting or acting, may depend on a small number of hours), we have chosen to use the concept of annual wage, the sum of pay received during the year in all jobs held by the employee. This requires us to set aside the intentionally low wages of individuals who work very little in the private sector over the course of a year (students’ summer jobs, etc.). We are keeping only those employees making more than half of the minimum wage annually. In the files, the concept of gross wage appeared better suited to representing wage-related phenomena (better coverage of different elements, greater stability over time). It does not, however, cover nonwage elements of compensation, such as stock options or allocation of shares, which leads us to underestimate a portion of the increase in inequality (Godechot 2012).
Aside from information about wage, number of hours, sector, and company, the databases contain information regarding the employee: sex, age, municipality of work and of residence (the latter only after 1993), and social category.
Let us highlight a few breaks in our datasets:
– First, in the exhaustive files, it is impossible to identify an employee from one job to another during the period from 1994 to 2001. Thus we are using the gross annual wages of full-time, nonsecondary jobs when these are greater than half of a minimum wage. This leads to setting aside some employees who worked multiple jobs during the year. The population is thus reduced by 16%. The exhaustive files of the year 2002, which also cover the year N-1, allow us to make a comparison between the two sets for the year 2001.
– Within the panel, the doubling of the covered population in 2002 (going from 1/24th to 1/12th of the workforce) may also introduce a modification of the representation of segregation phenomena measured on a small scale.
– Certain name changes also affect our study. The nomenclature of the Socioprofessional Categories (“Catégories Socioprofessionnelles” – CS) and the Socioprofessional Professions and Categories (“Professions et Catégories Socioprofessionnelles” – PCS) changed between 1982 and 1984 and between 2002 and 2003. Similarly, the sector names changed in 1993 and in 2008 (Nomenclature of Professional Activities [“Nomenclature d’Activités Professionnelles” – NAP], Nomenclature of French Activities [“Nomenclature d’Activités Française” – NAF] – 1993, then 2008).
Even though they do not cover the public sector and do not allow us to find nonwage income, the DADS are a privileged source for the study of wages with a significant level of detail and over a relatively long period of time. They make it possible to evaluate the evolution of the place of certain sectors in wage distribution.
15Figure 1 summarizes a certain number of results that we had established in our previous work dealing with the contribution of finance to wage inequality (Godechot 2012). The percentage of the full population of private-sector employees working in the financial sector has remained stable overall, and in fact has declined since the mid-1980s (from 3.5% to 3%). On the other hand, despite this overall stability, the higher one rises in the pay hierarchy, the greater the proportion of finance workers in the sample; this number has risen rapidly. Within the top ten percent, after a progression of one point at the beginning of the period, we observe the financial sector’s share stabilizing at about 8%. Within the top 1%, the growth of this sector has been steady, having doubled its share (from 6% to 12%) over the past twenty years. The slope is even more impressive when we concentrate on the top 0.1% of the nation’s earners with the highest wages. Within that group, the portion of finance workers has grown very rapidly, especially during the latter half of the 1990s, reaching 24%, or ten times their weight (in terms of odds ratio) in the rest of the wage distribution.
The share of finance employees among employees of different wage percentiles
The share of finance employees among employees of different wage percentilesNote: In 2007, 24.1% of employees in the top permille (0.1% of the highest paid) worked in finance.
16Wage distribution within the financial sector is considerably stretched out, to such a degree that it significantly transforms the level and the structure of inequalities. The wage share of the top 0.1% of earners rose from 1.2% to 2%, and nearly half of this increase went to the financial sector (Godechot 2012), a phenomenon similar to that which has taken place in the United Kingdom (Bell and Van Reenen 2010) and even in the United States (Bakija, Cole, and Heim 2010). All signs indicate that this movement is the result of the “financialization of finance,” or the increasing influence of the financial markets within financial and banking intermediation activities. It is strongly correlated with financial conditions (the mid-1980s boom, then the collapse following the 1987 crash and the Gulf War, a boom in the latter half of the 1990s followed by the 2002 recession, the mid-2000s boom) and with transaction volume (Godechot 2012), as well as with the significant weight of the socioprofessional professions and categories (PCS) “financial market executives” among the highest-paid finance workers (a classification measured only since 2003).
The Financialization of Nonfinancial Firms
17If the stretching of finance wages results, in a relatively obvious way, from the process of financialization and, consequently, represents it relatively well, it would nevertheless be incorrect to consider the effects of financialization as limited to the financial sector alone. Particularly in the United States, the ideological movement of “shareholder comeback” is manifested less by the mobilization of shareholders, as groups of natural persons, than by the development in the 1980s of mergers and acquisitions financed through leveraging (thanks to junk bonds) that made it possible for financial raiders to dismantle conglomerates (Zorn et al. 2005; Tomaskovic-Devey and Lin 2011, Lin and Tomaskovic-Devey 2013). In this threat environment, firms did not adopt all of the recommendations of financial governance from the financial theory of the firm, sometimes spread by management consulting firms (Lordon 2000; Froud and Williams 2000), but only those most directly related to short-term stock prices, in particular the promotion of financial directors to the rank of Chief Financial Officer (Deputy Director General) (Zorn et al. 2005) or the development of practices indexing directors’ compensation according to stock prices (Dobbin and Jung 2011). In the United States, one of the consequences of financialization is the growth in pay inequality and the rise of managerial wages (Lin and Tomaskovic-Devey 2013).
18Using the DADS, can one find evidence of the financialization of nonfinancial companies? The task is difficult, and the indicators that we have selected are rather mediocre. Drawing inspiration from Dirk M. Zorn’s approach (Zorn 2005; Zorn et al. 2005), we have attempted to see whether financial professions in nonfinancial companies have grown both in number and in special recognition. It is difficult to isolate the position of financial director as well as the previously mentioned jobs have been isolated. Based on the four-digit PCS, one might define the professions of executives in administrative, accounting, and financial services, among which financial executives, though not isolated, are the most numerous.  Additionally, the quality of the four-digit PCS, the nomenclature of which changed in 2003, is very poor in the DADS: it is all but unusable prior to 1997 and does not really improve until 2001, with less than one-third of earners having manifestly erroneous or missing codes. Despite the limitations of the data, the exercise produces the following results: an overall stability in the numbers of executives related to corporate finance and a slight growth in their share of the highest levels of wage distribution, particularly in the top 0.1% (Figure 2). Nevertheless, we will take a cautious view regarding any such change.
The share of employees in corporate finance among the top 0.1% of employees and the total population
The share of employees in corporate finance among the top 0.1% of employees and the total populationNote: In 2007, 14.1% of the top 0.1% of employees worked in the sector of firm administration (741J), with holdings. The change of nomenclature in the PCS in 2003 causes a break in the data.
19Faced with the limitations of the PCS to grasp this phenomenon, it is possible to return to the sector data, which are of better quality than the PCS in the DADS. Two sectors caught our attention in the 1993 NAF list. The first, 741J – Firm Administration, includes “holding activities”: to wit, the companies, generally with reduced personnel, within a group that exert primarily financial control over the other productive subsidies in the group. In 2007, according to a Liaison Financière (LIFI) study, those coded firms represented just 1.3% of wage-earners, but 38% of group heads. This method of organizing activity reflects certain characteristics of financialization: the primacy of financial indicators over industrial or commercial indicators. The second, 741G – Business and Management Consulting, only indirectly represents the financialization of non-financial companies. Often, these consulting firms are small or midsize companies whose internal operations are hardly financialized. They have nonetheless contributed, through the advice they dispense to large companies, to promoting a financialization strategy, as shown in studies on the proliferation of shareholder management systems (Lordon 2000; Froud and Williams 2000).
20Though they remain modest, these two sectors have increased their share in the overall workforce; the former grew from 0.8% to 1% between 1995 and 2007; the latter, from 0.5% to 0.9%. What is more, as evidence of the importance and value attributed to these (directly or indirectly) financial positions, their share in the top 0.1% of wages has grown even more aggressively, rising from 3.3% to 6.7% for consulting and from 9% to 14.4% for the firm administration sector. If the rise on the consulting side is more pronounced, that is more a reflection of the development of the sector as a whole. Holdings are advancing, according to a rise in their overrepresentation at the top of the pay hierarchy.
21Even if we do not identify the precise impact of financialization in nonfinancial firms in a very strict way, we have a set of indicators showing the growing recognition of the professions responsible for implementing it. This movement remains, however, lesser in scope than that which we have observed in the financial sector, which consequently justifies considering the latter as the archetypal sector of financialization.
2 – Financialization and the Global City
22What impact might financialization have on sociospatial inequality? Saskia Sassen, in her book on global cities (2001), looks at the question more broadly, studying the effects of globalization, of which she considers financialization to be a crucial element. At a time of globalized production of goods and services, the concept of a “global city” serves first and foremost, in her view, as a name for the places in which globally dispersed economic activity is coordinated – a coordination that simultaneously takes organizational, computerized, and financial forms (Sassen 2004). However, these new venues for coordination are not without ties to the traditional forms of territorial centralization.
23One might expect a decline in traditional major urban centers as a result of globalization. The drop in transportation costs and the global nature of markets, on the sides of both production and demand, could lead one to question the primacy of traditional major urban centers such as New York, London, Tokyo, or Paris, where it is particularly expensive to settle. In finance, the “drop in transportation costs” has taken an even more radical form. Securitization and dematerialization have radically liberated financial activity from its traditional locales. Some stars in the world of finance have taken advantage of this to set up their financial activity far from Wall Street, one example being Michael Milken, who established his business in the field of junk bonds in Los Angeles in the 1980s.
24Sassen explains in her book that the scattering of production made possible by globalization actually leads us to reassert the value of central job duties, and that their growing complexity requires increased demand for firms that specialize in providing services to other businesses (2001). Far from being destabilized, jobs in corporate finance and the services most needed by businesses (market finance, consulting, etc.) have become concentrated in the major urban centers. Furthermore, market finance itself is hardly threatened by the dematerialization of financial life. Face-to-face social contact and informal contact networks continue to play a major role, both as a medium for building a collective financial opinion and as a nexus that structures the labor market. This promotes the adhesion of market finance to a few centers (Sassen 2005). Furthermore, certain niches of financial activity, such as high-frequency algorithmic trading (Lenglet 2011), which performs brief microarbitrages of stock prices by means of automated transaction programs, are all the more profitable when located as close as possible to the computers of stock exchange institutions, which act as auctioneers and organize the finalization of transactions.
25From this analysis, one might deduce that the traditional financial capital of a country will be strengthened by the financialization process, rather than destabilized by new opportunities to break free from the territory.
26Figure 3 presents the evolution of the categories that (imperfectly) reflect corporate finance workers among the top 0.1% of earners in Île-de-France and in the French provinces, respectively. The overall level of financialization is higher in Île-de-France. However, while these professions are generally advancing in the top 0.1%, as previously indicated, as well as in their respective geographic areas, it is difficult to draw a conclusion regarding a difference in pace. If such a difference must be determined, it would be more in favor of the provinces, which seem to be making up for some lost time where corporate finance is concerned.
The share of employees in corporate finance among the top 0.1% of employees in the provinces and in Île-de-France
The share of employees in corporate finance among the top 0.1% of employees in the provinces and in Île-de-FranceNote: In 2007, 16.5% of employees in top 0.1% in Île-de-France worked in the firm administration sector (741J). 14.4% of the top 0.1% of employees in the provinces work in the same sector.
The change of nomenclature in the PCS in 2003 creates a break in the data. The distinction between the provinces and Île-de-France is based here on workplace. The proportion within the top 0.1% of each geographic region has been calculated and not within the total of both together. The threshold for membership in the top 0.1% thus differs for each of the two regions.
27There is a significant contrast with what is observed in the financial sector. Indeed, as shown in figure 4, finance workers’ share of the top 1% and the top 0.1% of wages in Île-de-France grew sharply, rising in the former case from 6% in 1976 to 21% in 2007 and in the latter case from 5% to 37%, while over the same period the same sector’s share of the highest echelons of provincial wages barely changed at all.
The share of employees in the finance sector among employees in the provinces or Île-de-France (total, top 1%, and top 0.1%)
The share of employees in the finance sector among employees in the provinces or Île-de-France (total, top 1%, and top 0.1%)Note: In 2007, 37 % of employees in the top 0.1% in Île-de-France worked in the finance sector.
The distinction between the provinces and Île-de-France is based here on workplace. The proportion within the top 0.1% of each geographic region has been calculated and not within the total of both together. The threshold for membership in the top 0.1% thus differs for each of the two regions.
28What is the relationship between this move toward financialization and the dynamic of inequality in each of these regions? In the provinces, the wage gap remained remarkably stable. The top 1% accounted for between 5% and 5.2% of total wages in the provinces, while the top 0.1% accounted for between 1% and 1.2% (Figure 5). In Île-de-France, on the other hand, inequality grew considerably. The top percentile’s share of wages went from 6% to nearly 9% between 1996 and 2007; the top 0.1% went from 1.3% to 2.7%; and the top 0.01% went from 0.27% to 0.82%. 55% of the increase in total wages for the top 1% in Île-de-France between those two dates, as well as 62% of the increase for the top 0.1% and 77% of the increase for the top 0.01%, went to employees of the financial sector.
Share of the top 1% and top 0.1% within the total wages of the provinces and the Île-de-France
Share of the top 1% and top 0.1% within the total wages of the provinces and the Île-de-FranceNote: In 2007, the top 1% of Île-de-France received 8.8% of the total wages of Île-de-France; the top 1% of the provinces received 5.4% of the total wages of the provinces. The proportions have been calculated separately for both regions. The threshold for membership in the top 1% and 0.1% thus differ between the two regions.
29We can therefore conclude that the rise in wage inequality in France since the mid-1990s (Landais 2008; Solard 2010; Godechot 2012) is largely due to the increase in inequality in Île-de-France, which in turn is caused in large part (two-thirds, approximately) by the rise in top wages in the financial sector – and thus by the “financialization of finance” – and to a smaller degree (more difficult to calculate, due to the imprecision of our indicators) to the financialization of nonfinancial companies.
30One consequence of this increase in high wages in Île-de-France is that the highest earners in the provinces are clearly falling behind by comparison. The pay distribution gap between Île-de-France and the provinces is, admittedly, growing overall (see Appendix). The average wage in Île-de-France has gone from 1.32 times the average wage in the provinces in 1970 to 1.44 times the average wage in the provinces in 2000. Nevertheless, this divergence is growing primarily at the top of the distribution; the difference in the threshold for the top 1% in the two areas has gone from 1.53 to 1.89 (Appendix). The portion of workers in the top-earning percentile who are private-sector employees working in the provinces dropped from 46% to 35% between 1976 and 2007. Looking at only the top 0.1%, the decline is even more pronounced, with those in the provinces falling from 41% to 21% (Figure 6). If we approach the question by place of residence rather than by place of work, the evolution, though somewhat less striking, remains pronounced. The portion of the top 0.1% residing in the provinces declined, from 40% to 25%. In light of this wage quantile and this rough spatial categorization of Île-de-France vs. the provinces, we have evidence of a rather marked phenomenon of spatial segregation.
Share of different fractiles living and working in the provinces
Share of different fractiles living and working in the provincesNote: In 2007, 77% of private-sector employees lived in the provinces, and 75% worked in the provinces; 25% of the top 0.1% highest paid private-sector employees lived in the provinces, and 21% worked there.
31On this macroscopic scale, this phenomenon of segregation results less from residential choices within small geographic units and a phenomenon of social avoidance than from the spatial localization of activity – namely, the concentration of financial market management and intervention activities – and from the unequal distribution of wages for activities unequally distributed throughout the country.
32Thus the concentration of high earners in Île-de-France must be put in perspective with the development of the business districts around Paris. In terms of places of work, Paris has a concentration of between 35% and 45% of the top 0.1% of earners throughout the period studied. The Hauts-de-Seine department, on the other hand, saw its share grow considerably, rising from 10% of the top 0.1% in 1976 to more than 30% in 2007. During the same period, the share for the municipalities of Puteaux and Courbevoie, in which the La Défense district is located, rose from 2% to 12% of the top-earning 0.1% of workers. In 2007, the top 0.1% of earners constituted 1.2% of the employees working in these two municipalities, one of the highest rates in the country, behind the 8th arrondissement of Paris and Neuilly (1.3%).
33These results lead us to confirm, but also to nuance, the hypotheses of Saskia Sassen (2001). Contemporary economic changes have indeed considerably strengthened the great megacities and the Parisian region in France, but this is due above all to the financialization of finance, far more than to the financialization of corporates, never mind globalization or the reorganization of production.
3 – The Evolution of Sociospatial Segregation
34Is the impact of the financialization manifested primarily in Île-de-France enough to significantly modify the degree of sociospatial segregation? Do the highest earners and the lowest earners live more separately from each other than they did before? There have been relatively few studies of this issue – the usual angle being the segregation of minorities, rather than socioeconomic segregation – and they are often contradictory.
35On the one hand, Saskia Sassen (2001) has shown that the rise in power of global cities has gone hand in hand with an increase in inequality between those cities and the rest of their countries, as well as a rise in inequality within those cities themselves. However, she does not propose any specific measure of spatial segregation. Sean F. Reardon and Kendra Bischoff (2011), using a study of the one hundred largest metropolitan areas in the United Sates, have shown a three-point increase (or a multiplication by 1.3) in their indicator of spatial segregation in those cities between 1970 and 2000, and that this rise in spatial segregation (by income rank) was as strong as the rise in absolute income inequality (measured by the Gini index). In particular, they show that a rise of one standard deviation in the income inequality indicator led to a rise of 25% of the standard deviation of the segregation index.
36In France, Éric Maurin (2004), based on a study of the social homogeneity of survey strata from the Employment Survey (100,000 people surveyed per year), found socioeconomic residential segregation stable in France between 1991 and 2002. Similarly, Nina Guyon (2012), using exhaustive tax data for Île-de-France between 2000 and 2009, drew a similar conclusion. All of the many segregation indicators used indicate overall stability in the level of segregation in Île-de-France; in greater detail, they reveal a noticeable rise between 2000 and 2002, stability from 2002 to 2008, and a drop between 2008 and 2009. On the other hand, based on the 1990 and 1999 census, Edmond Préteceille (2006) observed an increase in the segregation of private-sector executives and the liberal professions in Île-de-France during that period. These studies tend to remain very general in nature (an overall segregation index with a possible study of the top and bottom deciles) and do not isolate with sufficient precision the levels of income-related stratification most affected by financialization.
37We propose to examine this issue with the help of the DADS, which admittedly present some flaws, but also numerous advantages. Among the flaws, we are working with an incomplete population: private-sector employees. Many others who also contribute to the phenomenon of spatial segregation – public-sector employees, the self-employed, the unemployed, and those outside of the workforce – are missing. Next, the smallest geographic unit available to us is the municipality of work or of residence, the sizes of which are highly variable, and based on their dimensions, these will reflect more or less accurately a phenomenon of copresence or avoidance.  However, these flaws are counterbalanced by several advantages: a good description of wages and very good coverage of the population concerned, both in the 1/24th sample from 1976 and 2001 and in the 1/12th sample from 2002 to 2007, and even more so in the exhaustive files from 1994 to 2008.
38To calculate segregation, we are using the classic Duncan dissimilarity index: Dg = 0.5* ?i|ngi/ng. – n-gi/n-g.| where ngi represents the number of group g in geographic unit i and n-gi represents the number of other groups (-g representing groups other than g) in the same unit, while ng. and n-g. represent the number of g groups and of other groups in the total population. This aggregate index has a simple interpretation: it represents the portion of the population of group g that would have to be displaced in order to obtain an equiproportional distribution of the group in all territorial units. Based on this classic index, Sean F. Reardon and Glenn Firebaugh propose a multi-group dissimilarity index as a weighted sum of the indicators of each group: D = [?g (ng./n) * (1 – ng./n). Dg]/[?g (ng./n) * (1 – ng./n)]. We are dividing the population into five relative wage groups: F0-25, the lowest-paid quartile; F25-75, the two median quartiles; F75-90, the bottom portion of the top quartile; F90-99, the comfortable top decile (minus the top 1%); and F99-100, the top 1%. This definition does not depend on absolute wage level, but only on rank within the wage ladder. Therefore, an absolute rise in wage has no automatic impact on the spatial distribution of relative wages.
39Figure 7 presents the evolution of the multigroup dissimilarity index for the municipality of work and residence in France, Île-de-France, and the provinces, as well as the department of residence and work.  Above all, it presents a certain number of characteristics in level. In the provinces, workplaces are a bit more segregated than places of residence. We find rather the opposite phenomenon in Île-de-France. Furthermore, the level of residential segregation in Île-de-France is higher than in the provinces, while in terms of work segregation, the opposite seems to be the case. 
Synthetic indicators of spatial segregation
Synthetic indicators of spatial segregationNote: In 2008, the index of dissimilarity based on municipalities of work (exhaustive) rose to 21.9%. Thus 21.9% of the population would have to move their places of work for the composition of each municipality to be equivalent to the overall composition. Calculations for Île-de-France (and respectively for the provinces) measure spatial segregation in Île-de-France and are based on the salary level of persons working in Île-de-France.
40On the other hand, on initial examination, the graph does not present any particularly clear changes. The observation oscillates, depending on the sets and the indicators, between stability and a slight increase. Let us look at this in detail.
41In terms of place of work, the observation shows a slight increase in segregation. Based on the sample, we get a relatively clear rise in spatial segregation between 1976 and 1993. The overall dissimilarity index rises from 1.4 points (department) to 2 points (municipality), or multiplication by 1.12-1.13 in terms of odds ratio. What follows is a bit more ambiguous and will differ according to the selected set (sample or exhaustive). We will summarize this with overall stability between 1994 and 2001 and another rise (based on the exhaustive files) of about 2 points between 2001 and 2008, whether we are looking at the whole sample, Île-de-France, or the provinces (multiplication by 1.15). Even conservatively, we can therefore summarize this overall evolution with an observation of a slight rise in spatial segregation at work.
42In terms of place of residence, the overall observation is one of strong stability beginning in the mid-1990s. The dissimilarity rate stood at 16.65% in 1995 and at 16.74% in 2008, with the rate varying by hardly more than half a point at any time during this period. We largely confirm Eric Maurin’s analyses (2004). Nevertheless, let us first note that prior to this period, the index based on department of residence (due to the unavailability of the municipality of residence in the data) shows an increase similar to the one observed for the department of the workplace, for an increase of one point between 1976 and 1993. We must further note that the fate of spatial segregation is somewhat different in Île-de-France and in the provinces. In Île-de-France, the tendency is a rise in spatial segregation on the order of 1.2 points, while in the provinces, spatial segregation appears to have fallen on the order of half a point.  Thus we return to Nina Guyon’s observation (2004) of a clear rise in segregation between 2000 and 2002, followed by stability. 
43The overall observation, then, oscillates between stability and a slight increase in spatial segregation (for workplaces in general and for residences in Île-de-France). In any case, we observe no great upheaval and a milder rise than the phenomenon observed in the United States (Reardon and Bischoff 2011), where the increase in inequality has existed longer and is more sustained. Nevertheless, the overall indexes tend to add up uniformly phenomena that do not necessarily have the same meaning, such as, for example, a phenomenon of segregation or desegregation in the median or extreme quantiles.
44Figure 8 presents a breakdown of the aggregate dissimilarity index according to our five relative wage classifications. As others have previously shown (Maurin 2004; Préteceille 2006; Guyon 2012), the phenomenon of social concentration according to income level primarily affects high incomes rather than low. In 2008, in order to obtain a proportional distribution of different classifications in each municipality, you would need to displace at least 14% of the lowest quartile, 13% of the two median quartiles, 17% of the F75-90, 31% of the F90-99, and 50% of the highest percentile of wages. Of course, equiproportionality is very hypothetical and there is very little chance of coming across such an arrangement “by chance,” especially in small municipalities. We have therefore calculated (for the year 2007), based on twenty simulations, the deviation from equiproportionality that would result from the chance allocation of persons of different classes in the different municipalities (keeping their current size). The “randomized” Duncan indexes would be as follows: 2.8% for the F0-10, 2.5% for the F25-75, 3.4% for the F75-90, 4.3% for the F90-99, and 12.3% for the F99-100.  Even once we have determined the chance element of this separatism, the observation of a stronger separatism at the top of the hierarchy is not upset.
Breakdown of residential spatial segregation by wage classification
Breakdown of residential spatial segregation by wage classificationNote: In 2008, the index of dissimilarity based on municipalities of residence rose to 50% for F99-100, the percentile of the best paid employees. At least 50% of the population of the top 1% would have to move to obtain an equiproportional distribution.
45In fact, some municipalities have a strong concentration of individuals belonging to the top 1%. In 2007, among the municipalities that were home to over one hundred private-sector employees, the following have the strongest concentration of the working rich: Aigremont (24%), Saint-Nom-la-Bretèche (23%), Feucherolles (20%), Marnes-la-Coquettes (19%), Fourqueux (19%), Chavenay (18%), Mareil-Marly (17%), Neuilly-sur Seine (17%). For the most part, these municipalities are located in the western suburbs of Paris, in the Yvelines and Hauts-de-Seine departments, surrounded by wooded areas and golf courses. The top arrondissements of Paris, such as the 7th and the 16th (14%), are a bit farther down the list. The 7th, however, retains the top position, among municipalities with at least one thousand private-sector employees, in the category of percentage of residents in the top 0.1% (4%), ahead of Neuilly-sur-Seine (3.7%).
46Aside from these variations of segregation by wage level, figure 8 highlights their conflicting trends between 1995 and 2008. Earners in the lowest quartile, and even those in the middle quantiles (F75-90 in particular), tended to experience something of a desegregation, with the lowest quartile seeing a two-point drop in its index. The concentration rate of top-decile earners, on the other hand, rose a bit (by one index point), and the top percentile’s rate rose more sharply (3.5 additional points, or a multiplication by 1.15). The trends in the indicators of exposure make it possible to examine in detail what the wealthiest workers are exposed to.  On average, 58% of the workers living in the same municipality as these individuals were in the lowest three wage quartiles in 2008; they had represented 61% of the population in 1995. The wealthiest are increasingly exposed to the top 10% of earners: the rate rose from 17.3% to 18.2% for the F90-99 group and from 4% to 4.5% for the top 1%.
47Furthermore, in figure 9, we propose to study the evolution of the residential and workplace segregation of the top 1%, in France, Île-de-France, and the provinces. This analysis shows that, contrary to what we found in the aggregate index, residential segregation exceeds workplace segregation among the highest earners. Thus, in 2007, the city in which the largest number of top-percentile earners worked (among cities with over 1,000 employees) was Courbevoie, with a rate of 8%, which represents a far weaker concentration than the residential concentration of Saint-Nom-la-Bretèche (23%). Nevertheless, it is in the workplace that spatial concentration has advanced the most: 10 additional points and multiplication by 1.5 (based on municipality) or 1.6 (based on department) between 1976 and 2007, and 7 additional points and a multiplication by 1.3 between 1995 and 2008. During the same period of time, the residential separatism of the highest earners grew by only 3.5 points. The second finding is that the segregation of the wealthiest increased twice as much in Île-de-France as it did in the provinces. The residential dissimilarity rate in Île-de-France rose by 3 points, vs. 1.8 in the provinces; the workplace rate rose by 8 points, vs. 3 points in the provinces. The difference in heterogeneity has thus only grown.
Development of the separation of the top 1%
Development of the separation of the top 1%Note: In 2008, the dissimilarity index of Île-de-France top 1% based on municipality of residence (exhaustive) rose to 50%. Thus at least 50% of Île-de-France top 1% would have to move to achieve an equiproportional distribution in this region.
48Focusing on the highest-paid one percent of earners, then, one can conclude that there is a clear trend toward more segregation, especially in Île-de-France. This observation is consistent with both Saskia Sassen’s general framework regarding global cities and Edmond Préteceille’s analysis (2006) of the spatial segregation of private-sector executives in Île-de-France. It may appear to be out of step with Nina Guyon’s results (2012), which allowed her to conclude, based on an analysis of the top decile, that segregation is stabilizing at the top. However, her results, which are based on a different population, do not analyze the specific fortune of the members of the top 1%, and are affected in part by the 2009 recession.
4 – An Evaluation of the Impact of Financialization
49We have shown that, during the time when financialization was transforming the position and composition of the wage elite, they tended to live and to work (in more modest proportions) in a manner more separated from the rest of the other wage quantiles. This simultaneity is insufficient evidence of a cause-and-effect relationship.
50In an effort to evaluate this more specifically, we propose to carry out a counterfactual exercise: trying to calculate what the level of spatial segregation would have been in the absence of the financialization process. We will restrict this exercise to the financialization of the financial sector, since we have only poorly identified the financialization of nonfinancial companies. As an indicator of the financialization of finance, we will use one of its clear, direct consequences: the explosion of top wages and inequality within the financial sector. The counterfactual experiment thus consists of replacing the finance workers from 2007 with those from 1995, with their share of total wages and their municipalities of residence and work.  The difference between the real trend and the counterfactual trend will allow us to identify the impact of financialization. Let us note that, to the extent that the distortion of wages in the financial sector is an incomplete measure of financialization, it is likely that we are somewhat underestimating the sociospatial impact of financialization.
51Table 1 presents the results of such an analysis. As we have already demonstrated by another method (Godechot 2012), the rise in wages in finance has had a noticeable impact on the structure of inequality. Without it, the growth of the top percentile’s share would have been 26% lower, and the top 0.1%’s share 33% lower.
Simulation replacing finance employees of 2007 with those from 1995
Simulation replacing finance employees of 2007 with those from 1995Note: 38.2% of employees in the top percentile lived in the provinces in 1995; 35.6% lived there in 2007. This rate would have been 36.8 if the finance sector of 2007 had retained the same places of residence and work and the same salary amount as in 1995. The transformations of the finance sector thus have contributed to 45% of the development of the top 1% between 1995 and 2007.
52This effect on overall inequality is also expressed spatially. The Paris-province distribution of elite wages also would have been different. The decline of the provinces in the top 1% and (even more so) in the top 0.1% would have been less pronounced: the drop would have been reduced by 45% for the top 1% and by 70% for the top 0.1%. The increase, albeit a moderate one, in the residential segregation of the top earners would have been 40% less. Curiously, the clearer rise in workplace segregation seems less affected by the process of financialization, the impact of which is just 16%. This effect is perhaps the result of the complexity of the dynamic of professional establishment at the municipality level. On the other hand, once we look on a larger scale, such as the department, we get a more substantial contribution from financialization on the separation of the highest earners.
53To summarize this counterfactual analysis, then, we would say that the financialization of the financial sector, seen through the deformation of the wage structure in that sector, contributed to 20-40% of the rise in the top percentile’s sociospatial segregation and 40-70% of the segregation of the top 0.1% of earners in France.
55The first contribution of this article is to illustrate several trends regarding sociospatial inequality. The rise in high wages and wage inequality in France is centered mainly in Île-de-France, which has led to the disappearance of residents of the provinces from the ranks of the high-earning elite. Measured in the aggregate on the basis of municipality, residential sociospatial segregation has certainly remained stable from 1995 to 2008. On the other hand, the segregation of the best-compensated workers has indeed increased during the same period, especially when measured in Île-de-France alone or on the basis of workplaces.
56The second contribution is to isolate one of the determinants of this trend: financialization. This movement, with principally affects a small elite, is admittedly not fundamentally upending the aggregate measures of inequality or spatial segregation, but it nonetheless produces noticeable effects on wage inequality and spatial inequality surrounding the highest wages. Thus we show that it is responsible for at least 20-30% of their increase and that it is helping to widen the gap between Île-de-France and the provinces. One important point to continue this work would be to analyze the interaction between these trends and the differentiated price dynamics in the two subsets on the real estate market. 
57The gap between Paris and “the French desert” is a well-known and long-condemned phenomenon (Gravier 1947). The traditional reading of this phenomenon is above all political (the weight of monarchical and Jacobin tradition) and cultural (the centralization of cultural and journalistic life). Today, it is primarily the current economic and financial changes that are leading to a reinforcement of this imbalance. We are admittedly lacking a globalization indicator to separate what is due to financialization and what is due to nonfinancial globalization. Through the example of Paris, however, it seems that it is primarily financialization, rather than globalization, that is leading to the increased domination of the major megacities. Are not Paris, New York, London, and Tokyo primarily “financial cities,” rather than “global cities” (Sassen)? A comparative study of countries in which the economic, financial, political, and cultural functions of the great megacities are separated (Italy, the Netherlands, and Germany, for example) could shine a light on the foundations of their domination and its renewal.
Evolution of relationships between the thresholds of wage distribution in Île-de-France and in the provinces
Evolution of relationships between the thresholds of wage distribution in Île-de-France and in the provincesNote: In 2007, the threshold for P99 salaries in Île-de-France (the level above which the highest-paid 1% are found) was 1.9 times higher than the P99 in the provinces. P99: lower limit of the highest-paid 1%; P95: lower limit of the highest-paid 5%; P90: lower limit of the top decile; P10: upper limit of the lowest decile; Q1: upper limit of the first quartile; Q3: upper limit of the third quartile.
The author received access to this data from the CASD (Secure Remote Access Center) dedicated to researchers granted authorization based on the recommendation of the French Statistical Confidentiality Committee.
Sources: the datasets “7.202 – Compte des institutions financières (S12A) (En milliards d’euros)” and “7.101 – Compte des sociétés non financières (S11) (En milliards d’euros),” Institut national de la statistique et des études économiques, available at http://www.insee.fr/en/themes/theme.asp?theme=16&sous_theme=5.3. Nevertheless, it is to be noted that the concept of company profit is poorly represented by gross operating income, because it does not take financing costs or property income into consideration. This limitation is particularly true of financial companies.
We define the financial sector as the combination of financial intermediation (sector 65 in NAF 2003) and auxiliary financial services (sector 67.1 in NAF 2003). Insurance and the real estate sector are excluded, which is not the case in Krippner (2005) and Tomaskovic-Devey and Lin (2011a).
In our previous work (Godechot 2012), we discussed in greater detail the advantages and limitations of this source and the specific choices of wage definition.
In the 2003 PCS nomenclature, one finds the following professions: 372a, Executives Responsible for Economic, Financial, and Business Studies; 372b, Executives for Organization or Supervision of Administrative and Financial Services; 373a, Executives for Financial or Accounting Services in Large Companies; 373c, Executives for Financial or Accounting Services in Small and Midsize Companies.
In the case of large cities (Lyon, Paris, Marseille), however, we do have access to the arrondissements.
The results derived from the sample are not directly comparable to those from the exhaustive files, particularly for the municipalities. The population of small municipalities may be poorly represented due to sample sizes in the panel, which could potentially exaggerate the gaps in an equiproportional representation.
The change in the set in 2001, which makes it possible to take into account better those employees working more than one job during the year, leads to a significant jump in the index for the municipality of employment.
The department of residence is often uncertain in the 1994 data, particularly in Île-de-France. This led us to take 1995 as a point of reference. We are subtracting the jump linked to the change in sets in 2001 from our progression.
Let us note that Guyon’s observation of overall stability was based in part on the year 2009, a year of deep recession that affected some high incomes (due to the disappearance of bonuses in some sectors) and modified the spatial concentration of the very wealthy.
The standard deviations for these simulations are very low, ranging between 0.1% and 0.6%.
The exposure of group x to group y, xPy, is calculated as follows: xPy = ?i (nxi/nx.)*(nyi/n.i), where nxi and nyi represent, respectively, the numbers of groups x and y in the geographic unit i; nx. represents the total number of group x; and n.i represents the number of unit i (Massey and Denton 1988).
The number of finance workers was lower in 1995 than in 2007, due to the overall growth of the workforce and, especially, the difference in the sets in 2001 (the date from which it becomes possible to follow better a portion of multijob workers; see box). In order to obtain the financial population in 1995 (404,069) with numbers equivalent to the 2007 population (473,572), we have supplemented it with a sample of 69,503 individuals drawn at random from the 1995 population. Since in this counterfactual population the sum of finance workers’ share of total wages in 1995 and that of nonfinance workers in 2007 cannot add up to 100%, we are normalizing the shares by dividing by the latter sum.
Such a connection does not seem so simple at first glance. Indeed, a comparison of the price indexes of apartments in Paris and in cities of over ten thousand residents in the provinces since 1994 shows an increasing divergence, both during the late 1990s and, especially, since 2007. On the other hand, during the period from 2001 to 2006, the rate of price increases was slightly higher in the provinces. See “Séries longues : Indices Notaires – Insee des prix des logements anciens,” Institut national de la statistique et des études économiques, http://www.insee.fr/fr/indicateurs/ind96/20100225/sl.xls.