Autologistic Actor Attribute Model Analysis of Unemployment: Dual Importance of Who You Know and Where You Live

Galina Daraganova and Pip Pattison

Unemployment: Location and Connections

Persistent regional unemployment disparities have been characterized as a major cause of regional decay and impose significant costs on communities (Bill, 2005; Mitchell & Bill, 2004). Macroeconomic explanations for the persistence of unemployment often revolve around economic factors, including spatial changes in the skill requirements of jobs, migration of jobs to the suburbs, persistent demand constraints, wage differentials, low labor mobility and related structural impediments, and variations in the distribution of industries across space (see reviews, for example, in Ihlanfeldt and Sjoquist (1998) and Ramakrishnan and Cerisola (2004)). Outside traditional macroeconomic explanations of unemployment at the local area level (e.g., suburb), explanations draw on theories of residential segregation (Cheshire, Monastiriotis, & Sheppard, 2003; Hunter, 1996), which suggest that similar educational background and socioeconomic status along with housing market factors play a substantial role in determining how people are distributed across geographic space. Over time, these differences may become more pronounced as people sort further along lines of race and income (Bill, 2005). Cheshire et al. argued that where people live does not drive inequality but rather determines geographic location of inequality:

Where people live and the incidence of segregation and ultimately of exclusion, mainly reflects the increasing inequality of incomes. So if either the incidence of unemployment rises and/or if the distribution of earning becomes more unequal then social segregation intensifies ... the poor are not poor, isolated and excluded for the reason which makes them poor. They are not poor because of where they live; rather they live where they do because they are poor. (2003, 83-84)

Although the importance of economic reasons and residential sorting of the population cannot be denied, the argument that there is no independent or additional effect of where you live and who you know on the likelihood of being unemployed may be overstated. A number of economists have emphasized that neighborhoods are not static units but comprise people who constantly interact with each other in particular times and places (Akerlof, 1997; Bill, 2005; Calvto-Armengol, 2006; Durlauf, 2004; Topa, 2001) - a point also echoed more generally by Abbott (1997). At the same time, these people are influenced by the choices of those others (Durlauf & Young, 2001), and the employment-related choices and outcomes of any one individual may well be dependent on the employment-related choices and outcomes of other individuals. These dependencies might be driven by either spatial or network proximity, so that the behavior of proximal others affects the information received by any one individual arising from the knowledge or behavior of others. As a result, decisions by different individuals about labor market activity may interact to influence the distribution of employment outcomes across individuals and neighborhoods leading, potentially, to suboptimal outcomes that persist in equilibrium (Durlauf, 2004; Wilson, 1987). That network proximity might have significant implications on a global scale was clearly illustrated in a study by Granovetter (1973) on job finding. He showed that less than 20% of male professional, technical, and managerial workers in Newton and Boston, Massachusetts, suburbs had obtained their jobs by formal means, such as simply replying to an advertisement or working through an employment agency.

Although the role of neighbors (defined by some suitable spatial or network proximity metric) in the persistence of unemployment has been documented in a number of empirical studies (e.g., Beaman, 2008; Conley & Topa, 2006; Hedstrom, Kolm, & Aberg, 2003; Topa, 2001; Wahba & Zenou, 2005), none of these studies has conducted an analysis at the level of data on individuals and networks, nor has the interdependence of network structure and spatial locations been studied.

Hence, the principal goal for the current empirical study was to assess the associations between employment status, social network ties and spatial proximity between individuals. The main hypotheses were (1) a person would be more likely to be unemployed if she/he was not socially active (i.e., not connected to many others); (2) employment status of individuals would be positively related to their partners’ employment status and the employment status of their partners’ partners; and (3) a person would more likely be unemployed if he or she resided in a region of unemployed individuals. To examine these three hypotheses empirically, we simultaneously analyze the individual impact of social network processes and geographic proximity effects on the distribution of employment outcomes using the detailed survey data on individuals and their network members.

 
Source
< Prev   CONTENTS   Source   Next >