Using plant-level data from the Plant Capacity Utilization (PCU) Survey, we examine how manufacturing plants' use of temporary workers is associated with the nature of their output fluctuations and other plant characteristics. We find that plants tend to hire temporary workers when their output can be expected to fall, a result consistent with the notion that firms use temporary workers to reduce costs associated with dismissing permanent employees. In addition, we find that plants whose future output levels are subject to greater uncertainty tend to use more temporary workers. We also examine the effects of wage and benefit levels for permanent workers, unionization rates, turnover rates, seasonal factors, and plant size and age on the use of temporary workers; based on our results, we discuss various views of why firms use temporary workers.
-
Professional Employer Organizations: What Are They, Who Uses Them and Why Should We Care?
September 2010
Working Paper Number:
CES-10-22
More and more U.S. workers are counted as employees of firms that they do not actually work for. Among such workers are those who staffed by temporary help service (THS) agencies and leased employees who are on the payroll of professional employment organizations (PEOs) but work for PEOs' client firms. While several papers study firms' use of THS services, few examine firms' use of PEO services. In this article, we summarize PEOs' business practices and examine how the intensity of their use varies across industries, geographic areas, and establishment characteristics using both public and confidential data.
View Full
Paper PDF
-
Strong Employers and Weak Employees:
How Does Employer Concentration Affect Wages?
April 2018
Working Paper Number:
CES-18-15
We analyze the effect of local-level labor market concentration on wages. Using plant-level U.S. Census data over the period 1977'2009, we find that: (1) local-level employer concentration exhibits substantial cross-sectional and time-series variation and increases over time; (2) consistent with labor market monopsony power, there is a negative relation between local-level employer concentration and wages that is more pronounced at high levels of concentration and increases over time; (3) the negative relation between labor market concentration and wages is stronger when unionization rates are low; (4) the link between productivity growth and wage growth is stronger when labor markets are less concentrated; and (5) exposure to greater import competition from China (the 'China Shock') is associated with more concentrated labor markets. These five results emphasize the role of local-level labor market monopsonies in influencing firm wage-setting.
View Full
Paper PDF
-
Predictive Analytics and Organizational Architecture:
Plant-Level Evidence from Census Data
January 2019
Working Paper Number:
CES-19-02
We examine trends in the use of predictive analytics for a sample of more than 25,000 manufacturing plants using proprietary data from the US Census Bureau. Comparing 2010 and 2015, we find that use of predictive analytics has increased markedly, with the greatest use in younger plants, professionally-managed firms, more educated workforces, and stable industries. Decisions on data to be gathered originate from headquarters and are associated with less delegation of decision-making and more widespread awareness of quantitative targets among plant employees. Performance targets become more accurate, long-term oriented, and linked to company-wide performance, and management incentives strengthen, both in terms of monetary bonuses and career outcomes. Plants increasing predictive analytics become more efficient, with lower inventory, increased volume of shipments, narrower product mix, reduced management payroll and increased use of flexible and temporary employees. Results are robust to a specification based on increased government demand for data.
View Full
Paper PDF
-
Firms and Layoffs: The Impact of Unionization on Involuntary Job Loss
March 2003
Working Paper Number:
CES-03-09
This paper focuses on the impact of unionization on involuntary job loss using establishment data from the 1997 National Employer Survey (NES-II) and merging those data with contextual data at the industry level as well as with local labor market data. The estimated logit models included information on unionization rates and employment security provisions present in collective bargaining agreements as factors influencing layoff rates for individual establishments, controlling for establishment size, firm structure, use of non-regular employees, product/service demand and local employment. Results show that the impact of unionization is not significant except for (1) establishments that operate in the non-manufacturing sector; and (2) establishments operating in industries that have major collective bargaining agreements which contain moderate employment security provisions. Under those conditions, unionization decreases layoff rates; otherwise, unionization has no effect on layoff rates. These results provide some evidence that unions may have placed increased emphasis on employment security in order to protect members against involuntary job loss. This is in contrast to earlier studies which found a positive relationship between unionization and layoffs. In addition, establishments in Right-to-Work states have higher rates of involuntary job loss.
View Full
Paper PDF
-
Computer Networks and U.S. Manufacturing Plant Productivity: New Evidence from the CNUS Data
January 2002
Working Paper Number:
CES-02-01
How do computers affect productivity? Many recent studies argue that using information technology, particularly computers, is a significant source of U.S. productivity growth. The specific mechanism remains elusive. Detailed data on the use of computers and computer networks have been scarce. Plant-level data on the use of computer networks and electronic business processes in the manufacturing sector of the United States were collected for the first time in 1999. Using these data, we find strong links between labor productivity and the presence of computer networks. We find that average labor productivity is higher in plants with networks. Computer networks have a positive and significant effect on plant labor productivity after controlling for multiple factors of production and plant characteristics. Networks increase estimated labor productivity by roughly 5 percent, depending on model specification. Model specifications that account for endogenous computer networks also show a positive and significant relationship. Our work differs from others in several important aspects. First, ours is the first study that directly links the use of computer networks to labor productivity using plant-level data for the entire U.S. manufacturing sector. Second, we extend the existing model relating computers to productivity by including materials as an explicit factor input. Third, we test for possible endogeneity problems associated with the computer network variable.
View Full
Paper PDF
-
HUMAN CAPITAL LOSS IN CORPORATE BANKRUPTCY
July 2013
Working Paper Number:
CES-13-37
This paper quantifies the 'human costs of bankruptcy' by estimating employee wage losses induced by the bankruptcy filing of employers using employee-employer matched data from the U.S. Census Bureau's LEHD program. We find that employee wages begin to deteriorate one year prior to bankruptcy. One year after bankruptcy, the magnitude of the decline in annual wages is 30% of pre-bankruptcy wages. The decrease in wages persists (at least) for five years post-bankruptcy. The present value of wage losses summed up to five years after bankruptcy amounts to 29-49% of the average pre-bankruptcy market value of firm. Furthermore, we find that the ex-ante wage premium to compensate for the ex-post wage loss due to bankruptcy can be of similar magnitude with that of the tax benefits of debt.
View Full
Paper PDF
-
Why Are Plant Deaths Countercyclical: Reallocation Timing or Fragility?
November 2006
Working Paper Number:
CES-06-24
Because plant deaths destroy specific capital with large local economic impacts and potentially important macroeconmic effects, understanding the causes of deaths and, in particular, why they are concentrated in cyclical downturns, is important. The reallocationtiming hypothesis posits that plants suffering adverse permanent demand/productivity shocks delay shutdowns until cyclical downturns when plant capacity is less valuable, while the fragility hypothesis posits that shutdowns occur in downturns because the option value of maintaining the plant through low profitability periods is too small. I show that the effect that a plant's specific capital has on the timing of plant deaths differs across these two hypotheses and then use this insight to test the hypotheses' relative importance. I find that fragility is the dominant cause of the countercyclical behavior of plant deaths. This suggests that the endogenous destruction of capital is likely an important amplification and propagation mechanism for cyclical shocks and that stabilization policies have the benefit of reduced capital destruction.
View Full
Paper PDF
-
Employer Health Benefit Costs and Demand for Part-Time Labor
April 2009
Working Paper Number:
CES-09-08
The link between rising employer costs for health insurance benefits and demand for part-time workers is investigated using non-public data from the Medical Expenditure Panel Survey- Insurance Component (MEPS-IC). The MEPS-IC is a nationally representative, annual establishment survey from the Agency for Healthcare Research and Quality (AHRQ). Pooling the establishment level data from the MEPS-IC from 1996-2004 and matching with the Longitudinal Business Database and supplemental economic data from the Bureau of Labor Statistics, a reduced form model of the percent of total FTE employees working part-time is estimated. This is modeled as a function of the employer health insurance contribution, establishment characteristics, and state-level economic indicators. To account for potential endogeneity, health insurance expenditures are estimated using instrumental variables (IVs). The unit of analysis is establishments that offer health insurance to full-time employees but not part time employees. Conditional on establishments offering health insurance to full-time employees, a 1 percent increase in employer health insurance contributions results in a 3.7 percent increase in part-time employees working at establishments in the U.S.
View Full
Paper PDF
-
Computer Networks and Productivity Revisited: Does Plant Size Matter? Evidence and Implications
September 2010
Working Paper Number:
CES-10-25
Numerous studies have documented a positive association between information technology (IT) investments and business- and establishment-level productivity, but these studies usually pay sole or disporportionate attention to small- or medium-sized entities. In this paper, we revisit the evidence for manufacturing plants presented in Atrostic and Nguyen (2005) and show that the positive relationship between computer networks and labor productivity is only found among small- and medium-sized plants. Indeed, for larger plants the relationship is negative, and employment-weighted estimates indicate computer networks have a negative relationship with the productivity of employees, on average. These findings indicate that computer network investments may have an ambiguous relationship with aggregate labor productivity growth.
View Full
Paper PDF
-
Economic Factors Underlying the Unbundling of Advertising Agency Services
August 2009
Working Paper Number:
CES-09-15
This paper addresses a longstanding puzzle involving the unbundling of services that has occurred over more than two decades in the U.S. advertising agency industry: How can the shift from the bundling to the unbundling of services be explained and what accounts for the slow pace of change? Using a cost-based theoretical framework of bundling due to Evans and Salinger (2005, 2008), we develop a simple model of an advertising agency's decision to unbundle its services as a tradeoff between the fixed cost to the advertiser of establishing and maintaining a relationship with an advertising agency and pecuniary economies of scale available in providing media services. The results from an econometric analysis of cross-sectional and pooled data collected by the U.S. Census Bureau for quinquenial censuses conducted between 1982 and 2002 support the key predictions of the model. We find that advertising agency establishments are more likely to unbundle if they are large and diversified in their service offerings and are less likely to do so with increasing age and greater geographical scope. We also find a strong trend toward unbundling over time, a result that is partially explained by increases in media prices over time.
View Full
Paper PDF