Abstract American families have become less economically secure in recent decades, and this process accelerated during the 2008 financial crisis and its immediate aftermath. This study investigates how the crisis apportioned income precarity among families compared to pre-crisis years. We use the Survey of Consumer Finances and find that working families suffered the preponderance of income losses from the crisis, although the crisis shifted income losses towards more privileged working families. In fact, middle-income working families now have the same level of income precarity as the working poor, and families in the top income quintile continue to have elevated precarity levels. This result indicates that the middle class continues to bear a growing share of economic risk and that all working families are experiencing heightened insecurity in the post-crisis era. 1. Introduction One of the more salient features of the neo-liberal era in the USA is families’ significant exposure to economic risk (Hacker, 2006). Not only are market gyrations intensifying, but families also have fewer social protections. As Galbraith has recently argued, economic instability and inequality are mutually reinforcing processes (2012), and a growing literature seeks to unpack the precise relationship between inequality and socio-economic vulnerability (Bania and Leete, 2009; Sandoval et al., 2009; Dynan et al., 2012; Hacker et al., 2014). While this interdisciplinary research has produced key insights, important questions about the relationship between inequality and income precarity remain. First, the established literature typically limits itself to working families (e.g. Hacker et al., 2014) or does not distinguish between working and non-working families (e.g. Sandoval et al., 2009), although employment status is a critical predictor of the risk an income loss or the likelihood an income windfall. In this study, we explicitly compare and contrast working and non-working families’ divergent experiences of income precarity. Second, the literature on job insecurity typically describes precarity as inversely related to income (Kalleberg and Sorensen, 1979; Rehm et al., 2012; Hill, 2013). Yet, in the wake of the 2008 financial crisis, job displacement and income losses were particularly bad for middle-class workers (Bivens and Shierholtz, 2013). This suggests that precarity is not concentrated among the most disadvantaged workers anymore, although it is unclear whether this was only the case during acute crisis conditions or that the American economy has transitioned to a ‘new normal’ (Summers, 2013) of broadly shared precarity. In this article, we address these issues and map the contours of income precarity in the USA. Specifically, we investigate how a family’s socio-economic position conditions its experience of income precarity, and whether the 2008 financial crisis precipitated a change in the distribution of income precarity. With the Survey of Consumer Finances, we draw on a pooled cross-sectional sample from the 1995–2013 surveys. We provide two new measures of precarity: the risk of an income loss, and the risk of an income loss adjusted for the likelihood of an income windfall. These dynamic measures correct for limitations we see in the economic insecurity literature insofar as they allow us to investigate a fuller range of resource gains and losses and to make comparisons across the income distribution over time. Our results indicate working families bore most of the excess income losses generated by the 2008 financial crisis, and that the increase in income precarity was concentrated among relatively privileged working families: whereas working families in the lowest income quintile had distinctively high levels of income precarity in recent years, this is no longer the case. Thus, systemic instability translates into micro-level insecurity depending on exposure to the labor market, and for working families insecurity is climbing up the income ladder. Our findings shed new light on the changing American class structure, offer a broader way to think about income precarity, and answer Western et al’s call (2012) to understand the mechanisms behind precarity. 2. Social stratification and income mobility Stratification is the unending process of sorting individuals and families into different socio-economic positions. Although stratification research often focuses on static measures such as cross-sectional income inequality, social systems are inherently dynamic and households are always exposed to the possibility of moving up or down in the rankings. This dynamic view of stratification is implicit in the ‘life chances’ concept. However, in quantitative research this probabilistic aspect has often migrated to the opposite side of a regression equation. Economic position is typically defined in static terms and employed as a predictor of important life outcomes, such as health, children’s educational attainment, and so on. More recently, scholars have called for a return to dynamic approaches to stratification (Western et al., 2012). What distinguishes the current wave is a growing emphasis on risk and insecurity and not just upward mobility or status attainment (Kalleberg, 2011; Hacker et al., 2014). Moves in this direction are important because the American economy as a whole has become less stable. This is not to say that upward mobility is no longer relevant. Rather, we need to account for both gains and losses for a truly dynamic understanding of stratification. To begin, we discuss two common ways of conceptualizing income mobility. While both of these approaches can provide important insights, we argue for a new approach that explicitly captures the risk of income loss, both in the absolute sense and relative to the probability of an income windfall. 2.1 Income volatility The first conceptualization is income volatility. Income volatility refers to individual or family income changes over time, including both losses and windfalls. The theoretical inspiration for this literature is Milton Friedman’s (1957) Permanent Income Theory, which envisions that each individual has a lifetime earnings trajectory based on their human capital. Actual income may differ from permanent income in a given year, but this residual is viewed as ‘transitory’. Many income volatility researchers seek to isolate and estimate this temporary component (Gottschalk and Moffitt, 2009), while others focus on income changes in two different time periods (Dynan et al., 2012). In the empirical literature, there is no consensus on the best way to measure volatility, whether by months (Bania and Leete, 2009), consecutive years (Dynan et al., 2012), whether income should be adjusted for family size (Gittleman and Joyce, 1999), or whether researchers should only focus on large changes in income (Dahl et al., 2008). Perhaps because of this methodological diversity and the use of multiple data sources, different studies suggest different volatility trends. The balance of evidence supports the view that family income volatility has risen in recent years (Gittleman and Joyce, 1999; Bania and Leete, 2009, Dynan et al., 2012), although Dahl et al. (2008) find no such trend for large changes in income. There is a consensus, however, on the relationship between income volatility and income level. Families at the bottom of the income distribution tend to have lower income volatility than families in the middle (Nichols and Rehm, 2014) or families in the top quintile (Dahl et al., 2008). Although the income volatility literature succeeds on its own terms, it raises some questions for researchers specifically interested in income losses. One question is the distribution of losses among all families, not just workers or with a head family member of prime working age. The income volatility literature tends to focus on working-age adults, and so, the results only extend to workers and working families rather than the full population of households. This is partly by design, because the underlying theory focuses on earnings, but is also a matter of convenience because workers transitioning into and out of the workforce are more difficult to model. For example, Rohde et al. (2014) restrict their analysis from age 30 to 55, while Nichols and Rehm (2014) use a range of 25–60. The other major limitation in the income volatility literature is that volatility is directionally neutral; an income windfall and income shock would both register as volatility. Of course, all families have some risk of income loss and some possibility of an income windfall. However, there is no reason to expect a priori that the distributions of losses and windfalls are identical with respect to socio-economic status (Dynan et al., 2012; Western et al., 2012). Thus, we need to distinguish between losses and gains in order to fully understand the allocation of precarity. 2.2 Economic insecurity Unlike the income volatility literature, studies focusing on economic insecurity incorporate the broader population and pay special attention to economic downsides. Importantly, these studies emphasize economic hardships, rather than income losses per se. For example, Sandoval et al. (2009) define insecurity as a spell below the poverty level, and find that younger adults were more at risk of experiencing poverty in recent decades. This indicates that poverty is becoming more transitory but that the risk of poverty is becoming more widespread. Relatedly, Hacker et al. (2014) define insecurity as a 25% drop in income adjusted for any wealth buffer, debt and medical expenses. They plot this ‘Economic Security Index’ across a variety of dimensions, and find that insecurity has been steadily rising in recent decades and that insecurity is inversely related to income. These studies are important contributions to our understanding of insecurity but they still leave some questions unanswered because they exclude many income losses form their analysis. For example, a relatively small drop in income might place a family below the poverty level but not appear as a large enough percent loss to register on Hacker et al’s (2014) Economic Security Index. Likewise, a family might suffer a large percentage income decline, but this loss would not be counted by Sandoval et al. (2009) if it remained above the poverty line. Nor would it be counted in Hacker et al’s (2014) method if this family had several months’ worth of savings. This wealth criterion in the latter instance is well adapted to Hacker et al’s (2014) interest in hardship-inducing losses but tilts their metric towards the most disadvantaged: In 2013, 94.5% of families had some liquid asset, and the median value of total liquid assets for these families was $21,000 (Bricker et al., 2014). The result is that many income losses by relatively privileged working families remain unaccounted for in these analyses. A focus on hardship is obviously important, especially for policy-making. But to arrive at a general understanding of precarity, we need dynamic measures that capture people’s vacillating social position and which allow for comparison of risk exposures across the class system. 3. Income precarity and the ups and downs of unstable times The income volatility and economic insecurity literatures draw our attention to fluctuations and hardships but they do not give us a full sense of the ups and downs families face over time. We therefore put forth the concept of income precarity. Income precarity refers to downward changes in a family’s income and income stream. We measure income precarity in two ways: income precarity as probability of income loss, and income precarity as the probability of an income windfall minus the probability of an income loss. The first and simplest measurement of income precarity is the probability of income loss. This measure captures the risk of a family losing income in a given year. This is conceptually straightforward and allows us to compare experiences of income precarity among all families. Yet such a measure plays a surprisingly minimal role in the existing literatures on income volatility or economic insecurity. Perhaps this is because precarity is difficult to measure in most surveys. For example, if a family has higher income in year one compared to year two, we may be tempted to say that it suffered an income loss in year two. Yet there are other explanations, such as having experienced an unusual windfall in year one or having an income stream that varies year-on-year within a given range, as is often the case for jobs with a variable income component such as sales commissions. Having a predictably variable income is not the same thing as suffering an income loss from unemployment, illness, or family disruption. To focus on losses per se rather than instances of income returning to normal or year-on-year variability within a normal range, we employ a survey question that directly asks respondents whether and why income was uncharacteristically low in the last year. Therefore, we can tease out why losses occurred to more definitively measure unusual losses. The second measure is income precarity as a balance of risk and reward, a ‘risk-reward index’. We measure this as the probability of enjoying an income windfall minus the probability of suffering an income loss. If a family has a greater probability of enjoying a windfall than suffering an income loss, then their risk-reward index will be positive; if vice versa, then the family will have a negative risk-reward index value. This metric incorporates the recognition that every economic position has a built-in economic chute as well as an economic ladder, although for some positions the chute is especially wide and the ladder is especially narrow. The risk-reward index is especially important for studying the post-crisis era, when many commentators wonder whether certain privileged groups have been able to capture the upside during good times but shift losses during the bad times (e.g. Frank, 2011). These income precarity measures are designed to make comparisons across different classes. Many researchers in this area pay disproportionate attention to the most disadvantaged because they are said to be the most economically insecure (Kalleberg and Sorensen, 1979; Rehm et al., 2012). We do not dispute this—families with the lowest incomes certainly face profound uncertainty. But to fully understand precarity, we have to look at all classes for two reasons. First, as we argue above, existing measures are geared towards hardship-inducing losses (e.g. Hacker et al., 2014). Second, there are reasons to think that precarity is not only affecting America’s lower classes. Sociologists often discuss a bifurcated job structure in the USA in which there are well-paying, relatively secure jobs at the top and insecure, low-paying work at the bottom (Wright and Dwyer, 2003; Kelleberg, 2011; Dwyer, 2013). But other trends are making workers’ lives more unstable, and not just those in ‘bad jobs’. Non-standard work regimes such as temporary and contingent work are becoming common, even for skilled professions like academics and attorneys (Brooks, 2011). Shareholder value tactics promote flexible work regimes and downsizing efforts (Lazonick and O’Sullivan, 2000; Jung, 2015), and firms are frequently out-sourcing business functions, which also curtails stable employment, as does the decline of internal-labor markets (Hollister, 2004). Even for those holding ‘good jobs’, such jobs are prone to change with surprising frequency as economic instability worsens (Galbraith, 2012). These trends explain where heightened precarity is coming from and give a sense of how pervasive it may be. Thus, it is imperative we revaluate precarity across the class system in unstable times. 4. Income precarity and the financial crisis Stratification systems may be slow to evolve in normal times, but economic crises produce an unusual amount of economic losses that can dramatically shift mobility patterns. The 2008 financial crisis is an excellent example. With economic turmoil reaching levels not seen since the Great Depression, some families were destined to suffer from the fallout in the form of income losses. How these losses were distributed, however, depended upon which families were supported by state intervention and which families were left to the market. We argue that the US government responded differently to three economic constituencies. Broadly, the state protected investors through substantial market interventions, continued to assist families on fixed-incomes, and allowed working families to remain exposed to market forces. We argue that the different experiences of these three constituencies translated into different levels of precarity. In particular, we are interested in the broad swath of working families whose limited social protections exposed them to significant risk. Importantly, this group includes more privileged working families who have historically been less exposed to precarity. The most notable response of the American government to the crisis was a series of financial market interventions. These interventions were tilted towards the interest of investors and financiers rather than workers. For example, the Troubled Asset Relief Program (TARP) and the nationalization of Fannie Mae and Freddie Mac transformed trillions of dollars’ worth of questionable assets into the equivalent of risk-free government bonds (Acharya et al., 2011; Department of the Treasury, 2014, p. 3). As the financial system began to show signs of stabilization, the Federal Reserve continued its policy of purchasing bonds in order to increase the price of financial assets, further reducing risk for investors by ‘easing financial conditions’ (Bernanke, 2010). Although less dramatic, another pivotal decision by the American government was to continue providing income for millions of disabled and retired workers through the Social Security program. In 2014, the Social Security Administration’s (SSA) Old Age, Survivors, and Disability Insurance (OASDI) programs provided benefits to 41 million retired workers and dependents of retired workers, 6 million survivors of deceased workers, and 11 million disabled workers and dependents of disabled workers (Social Security Administration, 2014c). The SSA’s Supplemental Security Income (SSI) program, which is designed to assist aged, blind or disabled individuals with little or no Social Security or other household income, included more than 8 million recipients in 2014 (Social Security Administration, 2014a). This combined total of 66 million accounts is particularly significant when we consider that the U.S. has about 115 million total households (Vespa et al., 2013). However, the benefits themselves are relatively small. The average monthly Social Security benefit for retired workers in 2013 was $1,262, for disabled workers $1,130, while the average for SSI in 2014 was $516 per month (Social Security Administration, 2014a,b). Yet despite the fact that many disabled and retired individuals have relatively low levels of income, the state provides them with a more secure income stream than families depending upon wages for income. Unlike financiers, investors and families on a fixed income, workers were largely left to the mercy of the market during the crisis. While investors and families living on fixed-incomes may not share much in common, state assistance (however large or small), mutes the experience of market-driven precarity for both groups. Even before the crisis, working families faced heightened insecurity. In an average quarter from 1990 to 2005, 7.6% of jobs disappeared while 7.9% were totally new (Davis et al., 2006). This level of fluctuation is much higher than other events likely to trigger income losses such as childbirth, death, divorce, serious illness, or disability (Center for Disease Control, 2013a,b; Social Security Administration, 2014c). Once the crisis hit, the pace at which jobs disappeared accelerated while fewer new jobs became available. After the 2007 peak, 8.4 million jobs were lost. By the end of 2013, the American workforce was smaller than in 2007 by 1.6 million workers despite continued population growth (Katz, 2010; Bureau of Labor Statistics, 2014). We therefore expect that working families experienced the most of the excess income losses connected to the 2008 financial crisis: Hypothesis 1: Working families have higher levels of income precarity than those without workers, particularly since the 2008 economic crisis. If workers and their families absorbed the bulk of economic losses stemming from the crisis, which group of workers was affected the most? One possibility is that lower-income working families were most adversely affected because they are the most vulnerable to begin with (Gottschalk and Moffit, 2009). Indeed, prior research establishes a clear relationship between wage levels and job security: higher-paid workers tend to have greater job security (Kalleberg and Sorensen, 1979; Kalleberg, 2011; Rehm et al., 2012; Hill, 2013). Another is that a rising tide of instability lifted all boats equally. In this scenario, earnings risk would increase across the board but the relative level of risk exposure by income group would remain the same (Kalleberg, 2012, p. 434). A third possibility is that the crisis had a greater proportional effect on relatively privileged workers and their families. Prior research documents that higher-paying occupations realize premiums in part because they enjoy protection from market forces, either through licensing, unionization, credentialing, professionalization, or other forms of social closure (Weedon, 2002). Many of these jobs are governed by implicit contracts, and such unspoken promises may prove unreliable during economic turmoil (Hallock, 2009). Low-income workers were already exposed to hyper-competitive markets with few assurances of security before the crisis, and so the crisis may have muted effects on their experience of income precarity. Accordingly, the crisis may have transferred risk up the socio-economic ladder, placing middle-income earners in a more precarious position than before. Indeed, the weakness of the post-crisis recovery (Bivens and Shierholtz, 2013) points to a sustained, structural rise in downward mobility. In 2011, 45.5% of unemployed workers had been unemployed for more than six months. Middle wage earners were the most likely to face long-term unemployment. Those who did return to the workforce often experienced a downgrade because low paying jobs grew faster during the recession than other types of work (see Dufour and Orhangazi, 2014). We therefore expect that the relationship between income precarity and income is weakening: Hypothesis 2: The traditional inverse relationship between income precarity and income level among working families is weakening, particularly since the 2008 financial crisis. Along with a general weakening relationship between precarity and income, this trend may also be holding for educational attainment. Education is highly important for explaining earnings trajectories as well as economic security. Kalleberg (2011), for example, argues that highly educated workers are more likely to hold ‘good’ jobs and are therefore less exposed to precarity. Yet college graduates no longer appear immune from instability: whereas the unemployment rate for non-high school graduates increased 93% from 2007 to 2010 (9.7% to 18.8%), the unemployment rate for college graduates rose 146% during the same period (2.4% to 5.9%) (Economic Policy Institute, 2012, pp. 342–353). Since this time, unemployment and underemployment remains at elevated levels for college graduates while wage growth has stalled for this group (Kroeger et al., 2016). We therefore expect the following: Hypothesis 3: The traditional inverse relationship between income precarity and education among working families is weakening, particularly since the 2008 financial crisis. 5. Data and methods 5.1 Data To test these expectations, this study uses the Federal Reserve Board’s Survey of Consumer Finances (SCF). The SCF is a repeated cross-sectional survey conducted every three years by the Federal Reserve Board and includes a rich set of variables on household finance, including economic risk. The survey is a household survey, so respondents are asked about household-level finances rather than individual finances. Thus, it is perfectly suited for studying economic insecurity at the family level (Western et al., 2012). Males are recorded as the household head, although household heads are not necessarily the respondents interviewed (Lindamood et al., 2007). The sample contains an overweight of wealthy households (Kennickell and Woodburn, 1997. Survey weights are used in all the results presented here (Lindamood et al., 2007). The publicly available versions of the SCF are multiply imputed to protect the identity of the respondents and address any missing data. All estimates presented here also employ the repeated imputation inference method (Rubin, 1987). This study uses surveys from 1995 to 2013 because these years include all the indicators of interest. 5.2 Income precarity As indicated above, we measure income precarity two ways. Both are derived from a set of questions in the SCF that ask respondents whether income was unusual in the past year. After reporting last year’s income, respondents are then asked whether the amount they reported was normal, unusually low, or unusually high. The first income precarity indicator, the likelihood of loss, is based on the likelihood that a respondent reports unusually low income. We use logistic regression models to estimate family’s likelihood of an income loss in the previous year. The second is the risk-reward index—the estimated probability of an income windfall minus the estimated probability of an income loss. Values greater than zero on this metric indicate that windfalls are more likely than losses, while values less than zero indicate that losses are more likely than windfalls. To calculate this measure, we fit another series of logistic regressions which estimate the likelihood of a family experiencing an income windfall. We then subtract a family’s estimated probability of loss (our first measure) from the estimated probaiblity of a windfall. Measuring income precarity in these ways differs from related metrics used in prior research. First, compared to trajectory analyses that measure volatility, it focuses on income losses only, rather than variation around a long-term average (Gottschalk and Moffitt, 2009; Western et al., 2012). Second, it is different from Sandoval et al. (2009) analysis of poverty risk or Hacker et al’s (2014) Economic Security Index because it includes all losses rather than those below a certain income or wealth threshold. These hardship-related metrics are extremely useful, particularly in detecting especially damaging losses. But this study is concerned with income precarity across the socio-economic spectrum, and so does not impose restrictive conditions on which income losses are included. One limitation of this approach is that losses are only observed for a given year, although the effects of some losses take several years to dissipate (DiPrete and McManus, 2000). Unfortunately, the SCF data cannot address this issue, and longitudinal sources such as the Panel Study on Income Dynamics should be utilized to assess the durability of losses. Another limitation is that comparable SCF data on income risk is unavailable going back several decades, and so, this study is unable to determine whether risk levels in the 1990s and 2000s were higher compared to the pre-neoliberal era (Kalleberg, 2012). A third limitation is that the measures rely on respondents’ perceptions of income fluctuation. Small changes in income, whether gains or losses, may not be noticed, particularly if they are perceived as normal such as with cost of living increases. These small changes add up over the years and make a difference in a family’s ultimate economic position, although they imply a small year-on-year change in a family’s position. Given that we are interested in significant income movements in the short run, this issue does not pose much of a problem. However, this may affect risk-reward index values in ways we discuss in the results section. Despite these conceptual and methodological differences, there are important similarities with metrics used by other scholars. The percent of families reporting unusually low income due to earnings loss is 16% in the critical 2010 survey that asked about the immediate post-crisis period. This figure closely resembles the ‘over-the-year’ unemployment rate derived from CPS data as estimated by the Economic Policy Institute (2012, p. 355). The SCF also has an indicator for receiving unemployment income. This produces similar results as the income loss indicator. Thus, while this study employs these new metrics because they provide a holistic sense of precarity, they overlap in reassuring ways with other available measures.1 5.3 Labor force status, income level and education We code a family as being a working family if any family member has wage income or if the family suffered an income shock related to employment in the past year. If a family did not have wage income and suffered an income shock that was not an earnings shock—such as an investment loss, reduction in benefits or a change in family structure—that family was coded as not being a working family. To plot income precarity by income level, we compare loss incidence to the distribution of income that a respondent reports as ‘normal’. Normal income is defined as the income reported for a normal year and is derived from the set of questions described in the prior section. This is current income for families that report current income as normal, and income for a ‘normal year’ if current income is unusual. The goal of this indicator in the SCF is to approximate permanent income in a simple fashion that can be applied in a cross-sectional format (Bricker et al., 2014). This distinction between normal and current income is important: approximately 30% of households changed income quintiles in the 2010 survey, and we need to know the income of these families in a typical year prior to 2009 in order to observe how risk was distributed in 2009. Households are sorted into five normal income quintiles for all analyses. We code families by the education of the head with flag variables (which is assigned to males for married families) for bachelors degree and advanced degree along with similar variables for the household heads’ spouse. 5.4 Plan of analysis In order to estimate the probability of income loss and risk-reward ratio for families, we fit separate logistic regression models for income losses and income windfalls for each survey year and by labor force status. This results in 28 separate models.2 We estimate separate models because we (i) do not assume that the distribution of income losses is the same as the distribution of income windfalls (ii) expect that labor force status shapes the relationship between income and precarity (iii) want our modeling to reflect our expectation that precarity changes over time with economic conditions, particularly crises. In each model, we include control variables for demographic characteristics in order to account for factors which may simultaneously shape labor force status, income, education and precarity. We code household size as the number of individuals in the household, and age as the age of the household head. We also code the race of the household head with flag variables for black Hispanic, and other, with white as the reference group. After fitting each model, we calculate a family’s predicted probability of income loss or income windfall using the coefficients from the model. We then aggregate the predicted probabilities into subgroups as directed by our hypotheses (e.g. all working families in a given year, all working families in a given year in the 4th income quintile, etc.) in order to compare and contrast the average predicted probabilities of loss and risk-reward index values across different groups of families. Results are reported as the mean predicated probability for the subgroup with 95% confidence intervals. 6. Results 6.1 Descriptive statistics Table 1 summarizes demographic information for the entire sample and the subsamples of families with wage earners and those without wage earners. In the full sample, 18% of families reported an income loss. One-fifth (20.8%) of working families reported an income loss while 12.1% of families without wage earners reported an income loss. Thus, we see that families with wage earners in general have greater income precarity than those without. However, the descriptive table also shows that families with wage earners have significantly higher incomes, higher levels of educational attainment, larger households, higher marriage rates and have much younger household heads than families without wage earners. In fact, the average age of the household head for families without a wage earner is 66.8, indicating that this group includes a high proportion of elderly retirees. 44.4% of families without wage earners have a single female head (the SCF assigns household head status to males in married, heterosexual households), perhaps reflecting the fact that women comprise a larger share of the elderly because they tend to live longer than men. Table 1. Descriptive statistics for all households and working families, 1995–2013† All households Families with wage earners Families without wage earners Income shock 18.1% 20.8% 12.1% Mean household income $77,438 $87,340 $46,077 Median household income $45,360 $54,900 $20,988 Head has bachelor’s degree 17.8% 19.7% 11.6% Head has advanced degree 11.2% 12.1% 8.4% Mean household size 2.6 2.8 1.9 Head is married 51.3% 56.1% 36.1% Head is black 13.2% 12.9% 14.1% Head is Hispanic 8.7% 9.7% 5.3% Head is other non-white 4.0% 4.3% 2.7% Female head‡ 27.8% 22.6% 44.4% Mean age of head 49.6 44.1 66.8 N 34,480 26,380 8,100 All households Families with wage earners Families without wage earners Income shock 18.1% 20.8% 12.1% Mean household income $77,438 $87,340 $46,077 Median household income $45,360 $54,900 $20,988 Head has bachelor’s degree 17.8% 19.7% 11.6% Head has advanced degree 11.2% 12.1% 8.4% Mean household size 2.6 2.8 1.9 Head is married 51.3% 56.1% 36.1% Head is black 13.2% 12.9% 14.1% Head is Hispanic 8.7% 9.7% 5.3% Head is other non-white 4.0% 4.3% 2.7% Female head‡ 27.8% 22.6% 44.4% Mean age of head 49.6 44.1 66.8 N 34,480 26,380 8,100 † Survey weights employed. ‡ Figures are low because married households have males coded as head. Table 1. Descriptive statistics for all households and working families, 1995–2013† All households Families with wage earners Families without wage earners Income shock 18.1% 20.8% 12.1% Mean household income $77,438 $87,340 $46,077 Median household income $45,360 $54,900 $20,988 Head has bachelor’s degree 17.8% 19.7% 11.6% Head has advanced degree 11.2% 12.1% 8.4% Mean household size 2.6 2.8 1.9 Head is married 51.3% 56.1% 36.1% Head is black 13.2% 12.9% 14.1% Head is Hispanic 8.7% 9.7% 5.3% Head is other non-white 4.0% 4.3% 2.7% Female head‡ 27.8% 22.6% 44.4% Mean age of head 49.6 44.1 66.8 N 34,480 26,380 8,100 All households Families with wage earners Families without wage earners Income shock 18.1% 20.8% 12.1% Mean household income $77,438 $87,340 $46,077 Median household income $45,360 $54,900 $20,988 Head has bachelor’s degree 17.8% 19.7% 11.6% Head has advanced degree 11.2% 12.1% 8.4% Mean household size 2.6 2.8 1.9 Head is married 51.3% 56.1% 36.1% Head is black 13.2% 12.9% 14.1% Head is Hispanic 8.7% 9.7% 5.3% Head is other non-white 4.0% 4.3% 2.7% Female head‡ 27.8% 22.6% 44.4% Mean age of head 49.6 44.1 66.8 N 34,480 26,380 8,100 † Survey weights employed. ‡ Figures are low because married households have males coded as head. Table 2 documents the heterogeneity in economic status and risk profile among families without wage earners. Less than one-fifth of families in the bottom income quintile have investment income, compared to 57.4% in the middle quintile and 81.5% in the top quintile. Whereas investment income constituted 4.3% of the total for families in the bottom income quintile, investments generated 38.1% of total income for those in the top income quintile. In a supplemental analysis restricted to families without wage earners and a full set of demographic controls, families with income from investments had 49% lower relative odds of experiencing an income loss than those with no income-producing investments. These results underscore the diversity in economic situations among families without wage earners, particularly the insulation from income precarity enjoyed by families with investments. Although non-working families with investments are not the special focus of this study, future research should delve deeper into this group’s peculiar insulation from income precarity. Table 2. Descriptive statistics for families without wage earners, 1995–2013† Any investment income Investment income as percent of total Bottom quintile 17.8% 4.3% Second quintile 43.8% 8.7% Third quintile 57.4% 15.2% Fourth quintile 69.2% 20.5% Fifth quintile 81.5% 38.1% N 8,100 8,100 Any investment income Investment income as percent of total Bottom quintile 17.8% 4.3% Second quintile 43.8% 8.7% Third quintile 57.4% 15.2% Fourth quintile 69.2% 20.5% Fifth quintile 81.5% 38.1% N 8,100 8,100 † Survey weights employed. Table 2. Descriptive statistics for families without wage earners, 1995–2013† Any investment income Investment income as percent of total Bottom quintile 17.8% 4.3% Second quintile 43.8% 8.7% Third quintile 57.4% 15.2% Fourth quintile 69.2% 20.5% Fifth quintile 81.5% 38.1% N 8,100 8,100 Any investment income Investment income as percent of total Bottom quintile 17.8% 4.3% Second quintile 43.8% 8.7% Third quintile 57.4% 15.2% Fourth quintile 69.2% 20.5% Fifth quintile 81.5% 38.1% N 8,100 8,100 † Survey weights employed. To test our three hypotheses we compare the average estimated probability of income loss and risk-reward index across years and by labor force status. We combine the relevant regression results into readable figures that combine the results of several models, although the tables with regression coefficients are available in a supplementary. Figures 1 and 2 provide strong support for Hypothesis 1, which states that working families have higher levels of precarity than families without workers, especially after the crisis. Indeed, as Figure 1 shows, working families have a high probability of income loss in every sampled year. In 1995, for example, working families had a 20% likelihood of experiencing an income loss while non-working families were half as likely to experience an income loss. After the financial crisis, the probability that working families would experience an income loss jumped to 30% in 2010. For non-working families, this probability increased to 12%.3 We see similar patterns when looking at the risk-reward ration in Figure 2. Values below 0 indicate the probability of a net loss as compared with the previous year. All families had a probability of a net loss, but working families had less favorable risk-reward index values as compared with families without workers in every year. Importantly, we see that in 2010, working families’ risk-reward index fell to –0.23, meaning that income losses were much more likely than income windfalls. By contrast, the risk-reward index only fell to –0.08 for non-working families, indicating a much better balance of potential losses to potential gains. Figure 1. View largeDownload slide Average estimated probability of families experiencing income losses by labor force status and year. Estimates from logistic regression models with controls for education, income, and demographic characteristics (separate models were estimated by year and labor force status). Figure 1. View largeDownload slide Average estimated probability of families experiencing income losses by labor force status and year. Estimates from logistic regression models with controls for education, income, and demographic characteristics (separate models were estimated by year and labor force status). Figure 2. View largeDownload slide Average estimated probability of income windfall minus estimated probability of income loss for families by labor force status and year. Estimates from logistic regression models with controls for education, income, and demographic characteristics (separate models were estimated by year and labor force status for both income losses and income windfalls). Figure 2. View largeDownload slide Average estimated probability of income windfall minus estimated probability of income loss for families by labor force status and year. Estimates from logistic regression models with controls for education, income, and demographic characteristics (separate models were estimated by year and labor force status for both income losses and income windfalls). Figures 3 and 4 provide support for Hypothesis 2, which states that the typical inverse relationship between income precarity and income level among working families is weakening. Figure 3 shows the average estimated probability of an income loss by quintile. In early sample years, we can detect an inverse relationship between income and precarity. In 1995, for example, the bottom quintile had a much high probability of income loss (0.34) than did the middle quintile (0.23). However, this relationship is weakening. Even before the crisis—in the 2007 sample—we see that the bottom, second, and third quintile all had nearly identical average probabilities of income loss. The same is true after the crisis. In 2010, families in the fourth quintile had an average estimated probability of income loss comparable to their lower-income peers, with families from the bottom four quintiles hovering just above 0.30. The top income quintile consistently had a lower probability of experiencing an income loss across the period, although the risk steep gradient by income significantly eroded post-crisis compared to the 1990s.4 Figure 3. View largeDownload slide Average estimated probability of income loss among working families by income quintile and year. Estimates from logistic regression models with controls for education and demographic characteristics (separate models were estimated by year). Figure 3. View largeDownload slide Average estimated probability of income loss among working families by income quintile and year. Estimates from logistic regression models with controls for education and demographic characteristics (separate models were estimated by year). Figure 4. View largeDownload slide Average estimated probability of income windfall minus estimated probability of income loss by among working families income quintile and year. Estimates from logistic regression models with controls for education and demographic characteristics (separate models were estimated by year for both income losses and income windfalls). Figure 4. View largeDownload slide Average estimated probability of income windfall minus estimated probability of income loss by among working families income quintile and year. Estimates from logistic regression models with controls for education and demographic characteristics (separate models were estimated by year for both income losses and income windfalls). Figure 4 shows the risk-reward index estimates by quintile, and provides further support for the Hypothesis 2. Here, we see similar results as in Figure 3. In 1995, the bottom quintile had a lower risk-reward index value, indicating greater levels of precarity than higher-income workers. In 2004, the risk-reward index declined for all quintiles, indicating rising precarity across the board, although this trend was most pronounced in the middle. Just before the crisis in 2007, the second and third quintiles had the lowest average risk-reward index value. After the crisis, conditions continued to worsen. Notably, the estimates for 2010 increase from the first to the third quintiles (which has nearly identical levels as the fourth quintile at –0.24). That is, after the crisis, precarity actually increased with income except for the highest income quintile, indicating a partial reversal in the relationship between precarity and income among working families. Figures 5 and 6 provide support for Hypothesis 3, which states that the traditional inverse relationship between education level and precarity is weakening, especially after the crisis. Figure 5 shows the average estimated probability of income loss among working families by education level. In 1995, we see the inverse relationship between education and precarity as those with no degree had the highest probability of income loss (0.225), those with bachelor’s degrees had lower probabilities (0.17), and those with advanced degrees had the lowest probability (0.09). This gap persists in other sampling years, but in 2004 we see the probabilities of losses were similar for those with no degree and those with a bachelor’s degree. After the crisis, the probability of loss increased significantly for all groups, but there is still a gap between the three groups did not converge. Figure 5. View largeDownload slide Average estimated probability of income loss among working families by education status of household head and year. Estimates from logistic regression models with controls for income and demographic characteristics (separate models were estimated by year). Figure 5. View largeDownload slide Average estimated probability of income loss among working families by education status of household head and year. Estimates from logistic regression models with controls for income and demographic characteristics (separate models were estimated by year). Figure 6. View largeDownload slide Average estimated probability of income windfall minus estimated probability of income loss by among working families income quintile and year. Estimates from logistic regression models with controls for income and demographic characteristics (separate models were estimated by year for both income losses and income windfalls). Figure 6. View largeDownload slide Average estimated probability of income windfall minus estimated probability of income loss by among working families income quintile and year. Estimates from logistic regression models with controls for income and demographic characteristics (separate models were estimated by year for both income losses and income windfalls). Figure 6 shows mean estimated risk-reward index values estimates by education, providing weak support for hypothesis 3. For all years, we see a gap between education groups and their estimated likelihood of experiencing net losses. In 1995, those with no degree had the lowest risk-reward index value (–0.12), indicating the least favorable balance of income losses versus income windfalls. Those with advanced degrees were more likely to experience windfalls than losses in 1995, 2001 and 2007, as was also the case for those with a bachelor’s degree in 2001. However, the advantages of having a bachelors degree did not hold for all years. In 2004, those with a bachelor’s degree had a slightly higher average risk-reward index value (–0.14) as compared to those with no degree (–0.13). Thus, while there is evidence of growing precarity for the more educated, this was the only year this occurred. After the crisis, all groups experienced a much greater likelihood of experiencing a net loss—especially those with no degree and those with a bachelor’s—whose likelihood stood at –0.25 and –0.21, respectively. Thus, these estimates did become similar after the crisis. The average estimated risk-reward index value turned negative for the advanced degree group, although there remained a significant gap between these privileged families and those with less education. Taking Figures 5 and 6 together, we have support for hypotheses 3. Precarity levels are growing for all education levels, especially after the crisis. However, there are still gaps between the least and most educated families and the precarity gradient by education status has not totally flattened. 7. Discussion Americans are living in precarious times. While rising economic insecurity has been well documented by social scientists, we add to this literature by finding that that the extent of short-term income precarity is growing and affecting more Americans, including the middle class and the more educated. This trend was intensified by the 2008 financial crisis. While it is often said that economic precarity affects lower income families while well-off families are more secure (e.g. Rehm et al., 2012), we find that precarity has been creeping up the income ladder for two decades, and that this trend accelerated during the recession. Using pooled cross-sectional SCF data between 1995 and 2013, we devised two measures of income precarity: the probability of income loss and the probability of an income windfall minus the probability of income loss—a risk-reward index. We found that working families experience much higher levels of precarity than compared with non-working families—especially after the 2008 crisis. The former’s attachment to the labor market exposes them to the vicissitudes of that market while the latter is sheltered with (admittedly meager) state assistance. A key implication of this finding is that while the concepts of poverty and precarity are related, they are also distinct. Indeed, we also find precarity is rising to the middle and even upper-middle classes in the US. Even before the crisis, middle-income families experienced just as much or more precarity than lower income groups. Families across the income distribution experienced elevated levels of precarity following the crisis relative to the prior period, particularly the middle-income families. Finally, while we did not observe a convergence of income precarity across education levels, precarity is definitely climbing up the education ladder, increasing for all levels of education after the crisis. Taken together, our results indicate that economic precarity is not a unique experience to lower income families, but something more endemic to American families—a new reality that was intensified following the 2008 crisis. Our measures of income precarity are novel and differ significantly from existing measures of economic insecurity. Many economists measure income volatility while others look at the likelihood of facing economic hardships. While these approaches are important, they do not cover the whole population (income volatility), do not consider income gains along with losses, or do not include all types of income (economic insecurity). We address these concerns with new measures that allow us to make comparisons across the socio-economic spectrum. Of course, new measures raise methodological questions. SCF data is highly useful for our analysis, but it does not include certain characteristics that may be relevant. One important characteristic is occupation. Stratification scholars have emphasized occupation in shaping inequality (Wright and Dwyer, 2003; Weeden and Grusky, 2012; Dwyer, 2013) and occupational shifts are said to unevenly spread risk. Occupation is an important dimension to consider in relation to income precarity. For instance, occupation is central to Kalleberg (2011, p. 86), whose distinction of ‘good jobs’ and ‘bad jobs’ rests significantly on security. Yet, even he notes that ‘aside from U.S. Supreme Court justices and tenured faculty members, virtually all jobs are now more unstable as insecurity permeates the entire occupational structure. Even good jobs that pay well and provide opportunities for control and intrinsic rewards have become more insecure and stressful’. His contention is, of course, well in line with our findings and old lines between occupation and precarity are likely blurring. Nonetheless, occupation is an area worthy of future study. Another issue is that our measures of precarity are based on how respondents conceptualized ‘unusual’ income flows. The perception of what ‘unusual’ means varies by class, which could affect our results. Mullainathan and his colleagues, find that the less money an individual has, the better she is at gauging money losses and gains while the rich are prone to more errors (2013). This is not a serious threat to our study because we are primarily interested in precarity by labor force status and changes over time within income quintiles and education groups, and there is no evidence which indicates that the relationship between income and perception of losses has changed since the mid-1990s. However, it is important to note that income precarity estimates presented here may be unnecessarily low for the highest-income families due to underreporting. Our study of precarity adds to the discussion on stratification in the US, particularly the finding that median incomes have stagnated (Leicht and Fitzgerald, 2006). This combination of increased income precarity combined with cross-sectional wage stagnation could be playing out at the family level in a couple of different ways. It could be that individual workers no longer receive raises, or that workers continue to receive pay raises as they age, but compensation levels for workers at a certain age level remain frozen. In the latter scenario, all workers would individually enjoy upward mobility but overall incomes would never rise as higher-paid older workers retire and are replaced by lower-paid young workers. There is probably some of each process in the US stratification system. But this study highlights a third possibility: families increasingly suffer income losses and must struggle just to regain their prior economic position. This yearly churn forces many workers to restart their careers, making it harder to get ahead, and has become especially pronounced among middle-class families. In other words, tackling income stagnation will be even more difficult if middle-class families are finding it hard enough to tread water and downward mobility is a growing threat. Another issue is family-level risk pooling dynamics. This study does not delve into family structure beyond including risks from family disruption and other non-market risks. Of course, the institution of the family serves to lower risk exposure and pool risks across individuals, and further research should specify how family structure shapes risk profiles. For example, having two earners is often viewed as a mechanism to increase socio-economic status, but it might also increase income risk (Warren and Tyagi, 2003). The connection between having children or elderly family members and income risk is also unclear, as is the relationship between shrinking household sizes and the household risk distribution (Vespa et al., 2013). Moreover, extended family networks remain important for pooling both resources and liabilities and help to explain racial disparities in wealth and income (Heflin and Pattillo, 2006). Further research should focus on how these family dynamics interact with market-based risks. 8. Conclusion Rising income inequality is tied with rising income precarity. In this study, we have linked these two concepts by studying different groups’ precarity levels with a focus on the 2008 financial crisis. In doing so, we have made several contributions to the literature. First, we have developed a new and comprehensive measure of income precarity. Precarity is attracting more sociological attention, and researchers can use this piece as a springboard. Second, we have addressed Western et al.’s call (2012) for more ‘dynamic’ stratification models by considering income shocks and windfalls over time. Finally, we have advanced an understanding of the 2008 crisis and its fallout, which is attracting a great deal of interest vis-à-vis stratification (e.g. Pfeffer et al. 2013). The income risk distribution, as documented here, has important implications for both class analysis and theories of the welfare state. Although current income is a common empirical indicator of class or socio-economic status, different income streams contain different risks. In the past decade, the riskiness of wage income has increased, although other types of income have not. This indicates that the position of American working families has deteriorated to a greater extent than suggested by static measures of inequality and further confirms the growing domination of capital in the American stratification system (Kristal, 2010). One possible model for reform is the Danish policy of ‘flexicurity’, which maintains employer control over hiring and firing but combines this with a high minimum wage, an extensive array of social benefits for the unemployed and substantial investments in worker retraining (Kalleberg, 2012). Other possibilities include guaranteed minimum income, which is a basic income grant to households regardless of their participation in the labor force. Curiously, both progressives and conservatives have variously advocated for versions of this as a way to end poverty (Piven and Cloward, 1966; Friedman, 2002, pp. 191–194). On top of this positive effect, we note that it would have the added benefit of reducing income risk as well. Regardless of the specifics, reformers should address current risk disparities by both lowering systemic risk and by reallocating risk to households with the greatest capacity to absorb economic shocks. Supplementary material Supplementary material is available at Socio-Economic Review Journal online. Acknowledgements We would like to thank Rachel Dwyer, Randy Hodson, Andrew Martin and Vincent Roscigno for their thoughtful feedback. Footnotes 1 SCF data cannot be used to construct the Insecurity Index developed by Hacker et al. (2014) because it does not detail medical expenses or spells of poverty as described by Sandoval et al. (2009) because it does not track the same households over time. 2 These results are shown in supplementary tables. 3 In supplementary models that included all families, all years and a full set of controls, the risk disparity between working and non-working families widened since the financial crisis. Whereas in the pre-crisis period working families had 49% greater relative odds of experiencing an income shock than families without wage earners, this gap widened to 158% greater relative odds for the 2010–2013 period (difference between coefficients significant p < 0.001). 4 In supplementary models that included working families and a full set of controls, families in the bottom income quintile had 20% greater relative odds of income loss than top quintile families in the pre-crisis period but this risk gradient became non-significant in a model restricted to the post-crisis period (difference in coefficients across models significant at p < 0.05). References Acharya V., Richardson M., Van Nieuwerburgh S., White L. J. ( 2011) Guaranteed to Fail: Fannie Mae, Freddie Mac and the Debacle of Mortgage Finance , Princeton, NJ: Princeton University Press. Bania N., Leete L. ( 2009) ‘Monthly Household Income Volatility in the U.S., 1991/92 versus 2002/03’, Economics Bulletin , 29, 2100– 2112. Bernanke B. S. ( 2010) ‘Aiding the Economy: What the Fed Did and Why’, The Washington Post November 5, 2010. Accessed at http://www.federalreserve.gov/newsevents/other/o_bernanke20101105a.htm on November 25, 2014. Bivens J., Shierholz H. ( 2013) ‘The Great Recession’s Impact on Jobs, Wages, and Incomes’, In Wolfson M. H., Epstein G. A. (eds), The Handbook of the Political Economy of Financial Crises , New York, Oxford University Press, pp. 61– 95. Bricker J., Dettling L. J., Henriques A., Hsu J. W., Moore K. B., Sabelhaus J., Thompson J., Windle R. A., Devlin-Foltz S., Krimmel J. ( 2014) ‘Changes in U.S. Family Finances from 2010 to 2013: Evidence from the Survey of Consumer Finances’, Federal Reserve Bulletin , 100, 4. Brooks R. A. ( 2011) Cheaper by the Hour: Temporary Lawyers and the Deprofessionalization of the Law , Philadelphia, PA, Temple University Press. Bureau of Labor Statistics. ( 2014) ‘The Employment Situation – December 2013’,. accessed at http://www.bls.gov/news.release/archives/empsit_01102014.pdf on September 21, 2015. Center for Disease Control. ( 2013a) ‘National Marriage and Divorce Rate Trends’, accessed at http://www.cdc.gov/nchs/nvss/marriage_divorce_tables.htm on January 15, 2014. Center for Disease Control. ( 2013b) ‘Deaths: Final Data for 2010’, National Vital Statistics Reports , 61, 4. accessed at http://www.cdc.gov/nchs/data/nvsr/nvsr61/nvsr61_04.pdf on January 15, 2014. Dahl M., Thomas D., Jonathan A. S. ( 2008) Recent Trends in the Variability of Individual Earnings and Household Income . Washington, DC: Congressional Budget Office. Davis S. J., Faberman R. J., Haltiwanger J. ( 2006) ‘The Flow Approach to Labor Markets: New Data Sources and Micro-Macro Links’, The Journal of Economic Perspectives , 20, 3– 26. Google Scholar CrossRef Search ADS Department of the Treasury. ( 2014) Agency Financial Report: Office of Financial Stability – Troubles Asset Relief Program Fiscal Year 2014. accessed at http://www.treasury.gov/initiatives/financial-stability/reports/Documents/FY2014%20OFS%20AFR%20FINAL%20-%20 Nov %206%202014.pdf on November 25, 2014. DiPrete T. A., McManus P. ( 2000) ‘Family Change, Employment Transitions, and the Welfare State: Household Income Dynamics in the United States and Germany’, American Sociological Review , 65, 343– 370. Google Scholar CrossRef Search ADS Dufour M., Orhangazi O. ( 2014) ‘Capitalism, Crisis, and Class The United States Economy after the 2008 Financial Crisis’, Review of Radical Political Economics , 46, 461– 472. Google Scholar CrossRef Search ADS Dynan K., Elmendorf D., Sichel D. ( 2012) ‘The Evolution of Household Income Volatility’, The B.E. Journal of Economic Analysis & Policy , 12. Dwyer R. ( 2013) ‘The Care Economy? Gender, Economic Restructuring, and Job Polarization in the U.S. Labor Market’, American Sociological Review , 78, 390. Google Scholar CrossRef Search ADS Economic Policy Institute. ( 2012) The State of Working America , Washington, DC, Economic Policy Institute. Accessed at http://www.stateofworkingamerica.org/subjects/jobs/?reader on November 24, 2014. Frank R. ( 2011) ‘The Wild Ride of the 1%’, Wall Street Journal, October 22, 2011, Section C1. Friedman M. ( 1957) A Theory of the Consumption Function , Princeton, Princeton University Press. Friedman M. ( 2002) Capitalism and Freedom , Chicago, University of Chicago Press. Galbraith J. K. ( 2012) Inequality and Instability: A Study of the World Economy Just Before the Great Crisis , New York, Oxford University Press. Gittleman M., Mary J. ( 1999) ‘Have Family Income Mobility Patterns Changed?’, Demography , 36, 299– 314. Google Scholar CrossRef Search ADS PubMed Gottschalk P., Moffitt R. ( 2009) ‘The Rising Instability of U.S. Earnings’, Journal of Economic Perspectives , 23, 3– 24. Google Scholar CrossRef Search ADS Hacker J. S. ( 2006) The Great Risk Shift: The Assault on American Jobs, Families, Health Care, and Retirement , Oxford, Oxford University Press. Hacker J. S., Huber G. A., Nichols A., Rehm P., Schlesinger M., Valletta R., Craig S. ( 2014) ‘The Economic Security Index: A New Measure for Research and Policy Analysis’, Review of Income and Wealth , 60, 5– 32. Google Scholar CrossRef Search ADS Hallock K. F. ( 2009) ‘Job Loss and the Fraying of the Implicit Employment Contract’, Journal of Economic Perspectives , 23, 69– 93. Google Scholar CrossRef Search ADS Heflin C., Pattillo M. ( 2006) ‘Poverty in the Family: Race, Siblings, and Socioeconomic Heterogeneity’, Social Science Research , 35, 804– 882. Google Scholar CrossRef Search ADS Hill H. D. ( 2013) ‘Paid Sick Leave and Job Stability’, Work and Occupations , 40, 143– 173. Google Scholar CrossRef Search ADS Hollister M. N. ( 2004) ‘Does Firm Size Matter Anymore? The New Economy and Firm Size Wage Effect’, American Sociological Review , 69, 659– 679. Google Scholar CrossRef Search ADS Jung J. ( 2015) ‘Shareholder Value and Workforce Downsizing, 1981-2006’, Social Forces , 93, 1335– 1368. Google Scholar CrossRef Search ADS Kalleberg A., Sorensen A. ( 1979) ‘The Sociology of Labor Markets’, Annual Review of Sociology , 5, 351– 379. Google Scholar CrossRef Search ADS Kalleberg A. ( 2011) Good Jobs, Bad Jobs , New York, Russell Sage. Kalleberg A. ( 2012) ‘Job Quality and Precarious Work: Clarifications, Controversies and Challengers’, Work and Occupations , 39, 427– 448. Google Scholar CrossRef Search ADS Katz L. F. ( 2010) ‘Long-Term Unemployment in the Great Recession: Testimony for the Joint Economic Committee, U.S. Congress Hearing on ‘Long-Term Unemployment: Causes, Consequences, and Solutions’’, Cannon House Office Building, Room 210, April 29, 2010. Accessed at http://scholar.harvard.edu/files/lkatz/files/long_term_unemployment_in_the_great_recession.pdf on November 25, 2014. Kennickell A. B., Woodburn R. L. ( 1997) Consistent Weight Design for the 1989, 1992, and 1995 SCFs, and the Distribution of Wealth," Washington DC, Federal Reserve Board of Directors. Kristal T. ( 2010) ‘Good Times, Bad Times: Postwar Labor’s Share of National Income in Capitalist Democracies’, American Sociological Review , 75, 729– 763. Google Scholar CrossRef Search ADS Kroeger T., Cooke T., Gould E. ( 2016) The Class of 2016 , Washington DC, Economic Policy Institute. Lazonick W., O’Sullivan M. ( 2000) ‘Maximizing Shareholder Value: A New Ideology for Corporate Governance’, Economy and Society , 29, 13– 35. Google Scholar CrossRef Search ADS Leicht K. T., Fitzgerald S. T. ( 2006) Postindustrial Peasants: The Illusion of Middle-Class Prosperity , New York, Worth. Lindamood S., Hanna S. D., Bi L. ( 2007) ‘Using the Survey of Consumer Finances: Some Methodological Considerations and Issues’, Journal of Consumer Affairs , 41, 195– 222. Google Scholar CrossRef Search ADS Mullainathan S., Shafir E. ( 2013) Scarcity: Why Having Too Little Means So Much , New York, New York Times Books. Nichols A., Rehm P. ( 2014) ‘Income Risk in 30 Countries’, Review of Income and Wealth , 60, S98– S116. Google Scholar CrossRef Search ADS Pfeffer F. T., Danziger S. H., Schoeni R. F. ( 2013) ‘Wealth Disparities Before and After the Great Recession’, Annals of the American Academy of Political and Social Science , 650, 98– 123. Google Scholar CrossRef Search ADS PubMed Piven F. F., Cloward R. A. ( 1966) ‘The Weight of the Poor: A Strategy to End Poverty’, The Nation May 22, 1966. Rehm P., Hacker J. S., Schlesinger M. ( 2012) ‘Insecure Alliances: Risk, Inequality, and Support for the Welfare State’, American Political Science Review , 2, 386– 406. Google Scholar CrossRef Search ADS Rohde N, Tang K. K., Prasada Rao D. S. ( 2014.) ‘Distributional Characteristics of Income Insecurity in the U.S., Germany and Britain’, Review of Income and Wealth , 60, S159– S176. Google Scholar CrossRef Search ADS Rubin D. B. ( 1987) Multiple Imputation for Nonresponse in Surveys , New York, John Wiley & Sons. Sandoval D. A., Rank M. R., Hirschl T. A. ( 2009) ‘The Increasing Risk of Poverty Across the American Life Course’, Demography , 46, 717– 737. Google Scholar CrossRef Search ADS PubMed Social Security Administration. ( 2014a) Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Federal Disability Insurance Trust Funds. Accessed at http://www.ssa.gov/oact/tr/2014/tr2014.pdf on November 24, 2014. Social Security Administration. ( 2014b) Annual Report of the Supplemental Security Income Program. Accessed at http://www.ssa.gov/oact/ssir/SSI14/ssi2014.pdf on November 24, 2014. Social Security Administration. ( 2014c) Annual Statistical Report on the Social Security Disability Insurance Program, 2013. Accessed at http://www.ssa.gov/policy/docs/statcomps/di_asr/2013/di_asr13.pdf on January 15, 2014. Summers L. ( 2013) ‘Why Stagnation Might Prove to be the New Normal’, Financial Times . December 15, 2013. Vespa J., Lewis J. M., Kreider R. M. ( 2013) America’s Families and Living Arrangements: 2012 . U.S. Bureau of the Census. Accessed at http://www.census.gov/prod/2013pubs/p20-570.pdf on November 24, 2014. Warren E., Tyagi A. W. ( 2003) The Two Income Trap: Why Middle- Class Mothers and Fathers are Going Broke , New York, Basic Books. Weedon K. A. ( 2002) ‘Why Do Some Occupations Pay More than Others? Social Closure and Earnings Inequality in the United States’, American Journal of Sociology , 108, 55– 101. Google Scholar CrossRef Search ADS Weeden K. A., Grusky D. R. ( 2012) ‘The Three Worlds of Inequality’ American Journal of Sociology , 117, 1723– 1795. Google Scholar CrossRef Search ADS Western B., Bloome D., Sosnaud B., Tach L. ( 2012) ‘Economic Insecurity and Social Stratification’, Annual Review of Sociology , 38, 341– 359. Google Scholar CrossRef Search ADS Wright E. O., Dwyer R. ( 2003) ‘The Patterns of Job Expansions in the US: A Comparison of the 1960s and 1990s’, Socio-Economic Review , 1, 289– 325. Google Scholar CrossRef Search ADS © The Author 2017. Published by Oxford University Press and the Society for the Advancement of Socio-Economics. All rights reserved. For Permissions, please email: email@example.com
Socio-Economic Review – Oxford University Press
Published: Jun 21, 2017
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera