Almost Famous: How Wealth Shocks Impact Career Choices

Almost Famous: How Wealth Shocks Impact Career Choices

Jacelly Cespedes Zack Liu Carlos Parra?

January 2020

Abstract We study the short- and long-term career effects of shocks to household balance sheets using a novel dataset of workers in the film industry in the wake of the Great Recession. We find that individuals within the same county and occupation who lost more housing wealth reduce their participation in films with other high-profile talents (individuals that have won prestigious awards), films that are positively rated, and films that are likely to win awards, but they increase involvement in small productions. We control for local labor demand shocks by comparing homeowners to renters. These wealth losses also have adverse long-term consequences on the probability of leading roles, individuals' popularity, and film quality. Overall, our results suggest that wealth shocks distort non-salaried workers' labor decisions due to liquidity concerns and impact individuals' career paths.

We are grateful to Briana Chang, Murray Frank, Jeanne Lafortune, Borja Larrain, Jordan Nickerson, Richard Thakor, Tracy Wang, and seminar participants at the UT Dallas Finance Conference, 2019 ECWFC meeting at the WFA, PUC Chile, Universidad Adolfo Ibanez, University of Houston, and University of Minnesota for their helpful comments. Christos Kamaras and Alexandre Reggi Pecora provided excellent research assistance.

University of Minnesota. E-mail: cespe013@umn.edu University of Houston. E-mail: zliu@bauer.uh.edu ?Pontificia Universidad Catolica de Chile. E-mail: parra.c.carlos.r@

The collapse in house prices during the Great Recession led to an unprecedented decline in household wealth (Mian et al., 2013). While the literature has explored how housing-related wealth shocks affect labor supply, on-the-job productivity, and labor mobility (Brown and Matsa, 2016; Bernstein, 2017; Bernstein et al., 2018; Gopalan et al., 2019), it is also important to consider how these shocks affect other margins of adjustment, such as job characteristics and match quality. In standard job search models, wealth is a 'safety net' that allows individuals to be more selective during their search and to wait for better job opportunities (Shimer and Werning, 2007; Chetty, 2008). Thus, when agents experience a decrease in their housing wealth, which can reduce their reservation wages, they may take on job opportunities that do not fit their career objectives. These distortions to labor choices may have had aggregate effects on economic growth and recovery after the Great Recession.

We investigate what impact, if any, the deterioration of the household balance sheet exerts on individuals' career decisions and trajectories using the labor market for workers in the film industry. Specifically, we ask if housing wealth losses affect the types, quantity, and quality of films individuals choose to work on in the short and long term. Empirically estimating the effects of wealth shocks on labor market decisions is complicated by a few important issues, which we address in this paper. First, it is difficult to find detailed information on employment choices, especially data covering job quality and long-term horizons. Second, a comparison between financial health and career choices could be biased due, for example, to unobserved variables that can affect individuals' wealth, such as health shocks (e.g., Himmelstein et al., 2005; Gross and Notowidigdo, 2011; Ramsey et al., 2013). Third, it can be difficult to isolate the effects of house price declines from local labor market conditions.

In this paper, we compile a novel dataset of the job histories and biographical, and financial information for actors, directors, and writers in order to estimate the impact of wealth shocks on their subsequent labor choices following the housing crisis. Our empirical strategy exploits plausibly exogenous shocks to individuals' wealth, proxied by house price changes.1 In particular, we employ two empirical designs to identify how workers respond to these shocks. First, we exploit

1For example, Cheng et al. (2014) show that midlevel managers in securitized finance were unaware of problems in housing markets.

1

differences in homeowners' exposure to ZIP code house price changes within counties and within occupations.2 Second, to account for changes in the film industry labor market, we employ a difference-in-differences estimator that compares how homeowners and renters have responded to local house price changes following the Great Recession.3

The film industry is a good laboratory for examining the effects of wealth shocks on career decisions. First, employment in the film industry is primarily on a project-by-project basis, which requires workers to constantly search for employment opportunities. Additionally, since they are not salaried employees, housing wealth shocks might have outsized impact on their career choices. Second, we can collect detailed information about the quality and characteristics of projects, such as cast and crew, ratings, awards, and type of role. Thus, it is possible to categorize each job that individuals take. Third, we are able to hand-collect biographical information about workers and match it with other databases to determine the ZIP code of residence, homeownership status, and thus, exposure to local house price changes. Altogether, this allows us to examine the impact of house price shocks on labor decisions and also to look at long-run effects on workers' careers ten years afterward.

In this paper, we focus on film industry workers' participation in productions with different characteristics that would be known around the time the individual agrees to work on the film. We find that ex-ante job characteristics like production budget, crew size, and the talent of the cast are strongly correlated with box office returns and movie performance. In line with this, our main variables of interest are the number of films with high-profile cast and crew, the number of films with a large crew count, and the number of films with a small crew count, as proxies for big-budget and low-budget films.4 We define high-profile cast members as those who have won (or have been nominated for) Oscars or Golden Globes. While most individuals prefer to work on larger budget films that are likely to be box office hits and award-winning films, they may need to work on smaller productions to avoid financial distress.

2Bernstein et al. (2018), Pool et al. (2018), and Dimmock et al. (2018) use a similar empirical strategy to estimate the effect of changes in house price on risk-taking by firms' professionals around the housing crisis.

3Schmalz et al. (2017) employ an analogous difference-in-differences design to estimate the effect of house price appreciation on entrepreneurship.

4In general, the advantage of film projects with larger budgets is that they can afford more talented cast and crew members, more prep time before shooting begins, and more shooting days on set.

2

In our baseline specification, we compare the labor market decisions of individuals within the same occupation to those who own a house in the same county but who experience differential price declines in their respective ZIP codes. Our detailed data allow us to exploit the large variation in house price changes following the housing crisis within counties. In addition, we include controls for observable worker traits such as gender and age. We find that differences in ZIP-code-level house price changes significantly affect the choice of film projects. In particular, workers who lose more housing wealth are less likely to pursue high-profile films. In addition, individuals who experience adverse shocks also increase their participation in low-budget productions, and they are less likely to accept jobs in big-budget films. Our results show that home price shocks affect the ex-ante characteristics of the jobs that workers choose.

The decision to work on films with ex-ante low-quality might translate to low ex-post performance, potentially reducing the future opportunities for these individuals. Specifically, we find that house price losses negatively impact movie ratings, lower participation in films that win major awards, and decrease the likelihood of leading roles.

Subsequently, we analyze the long-term effects of wealth shocks. In the standard career model, where the labor market functions like a competitive spot market, any effects of bad luck are temporary and inconsequential for future career development (e.g., Mincer et al., 1974). In contrast, alternative models based on job search (Topel and Ward, 1992) and contracting (Harris and Holmstrom, 1982; MacLeod and Malcomson, 1993) rationalize how temporary shocks can have persistent negative effects on career outcomes. We find that housing wealth shocks have persistent, long-term effects on an individuals' popularity, on the probability of them landing leading roles, and of their projects having high film quality as much as 10 years after the onset of the Great Recession. These results highlight the potential disruption of career trajectories caused by housing wealth losses.

One potential concern is that a significant portion of the sample resides in Los Angeles County, where Hollywood is located, so a shortage of employment opportunities in the film industry during the recession could both lead to a decline in local house prices and affect workers' project selection.5 To address this concern, our second approach is a difference-in-differences strategy in which we

5Other counties with an important concentration of workers in the film industry are New York County (Manhattan), Orange County, and Kings County (Brooklyn).

3

expand our dataset to include renters in the same local labor markets. In this case, we compare the differential change in labor market decisions for homeowners and renters following the house price changes at the ZIP code level. The underlying assumption is that renters are a plausible control group who are exposed to the same film industry conditions but whose wealth is not as affected by changes in local house prices. Indeed, we find that only homeowners, not renters, change their film project decisions in response to local house price changes. This result confirms that the previous findings are not driven by an unobserved shock to the entire film industry.

Another concern is that changes in housing wealth do not affect all homeowners equally. While home price shocks serve as a good proxy for household wealth for most individuals, for very well-off workers, however, housing wealth might not represent a large portion of their total wealth. Thus, we would not expect house price changes to significantly affect the labor choices for these wealthy individuals. To proxy for wealthy workers, we classify those who live in ZIP codes where the average house price is greater than $2 million and those who own several properties. We find that housing wealth shocks have no effect on the labor decisions for wealthier individuals using these proxies.

In addition, we conduct a range of different checks to demonstrate the robustness of our results. First, we show that the results are robust to using alternative definitions of high-profile, small, and large films. Second, we examine a series of falsification tests to check whether outcomes measured during the 2003-2006 period are related to the individuals' exposure to house price changes around the housing crisis. We find that none of the outcome variables before the crisis can be explained by the house price changes during and after the crisis. Subsequently, we estimate an alternative model specifically for count observations. We present the analogs of our main specifications using Poisson fixed effect regressions, which corroborate the robustness of our main findings. Finally, we explore the degree of selection on unobservables needed to explain away the estimates using a partial identification approach (Altonji et al. 2005; Oster 2013). We find that the identified set for the different outcomes does not include zero.

The effect of a loss in housing wealth could affect the worker along several dimensions. First, many households were liquidity constrained during the financial crisis (e.g., lower opportunities

4

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download