© Copyright 2003 by the Wyoming Department of Employment, Research & Planning
Do Benefits Reduce Employee Turnover Among Wyoming Firms: A Response to the Workforce Development Council
by: Mark Harris, Sociologist, Ph.D.
"Benefits are a more powerful tool for reducing turnover among firms employing predominately part-time employees."
Benefits are often touted as a means to reduce employee turnover. Turnover reduction may be a goal for some employers to reduce replacement and training costs, increase productivity, lessen supervisory workload and stress, enhance customer service, or improve product
In this article, Research & Planning (R&P) seeks to determine whether offering more benefits reduces employee turnover among Wyoming firms. Results indicate that among the industry groupings examined, firms that offer more benefits have lower employee turnover. Additionally, the effect appears somewhat stronger in firms which employ predominantly part-time workers compared to those which employ predominantly full-time workers.
Data for this article come from three sources. Benefits data are from a random sample of 1,600 Wyoming firms as part of the 2000 Benefits Survey.2 The Benefits Survey reports whether sampled firms offer various benefits to their full- and part-time employees.3 Turnover rates4 (i.e., employee exit rates) are calculated using Wyoming Wage Records data from the same firms. Other characteristics of sampled firms (e.g., firm size, average quarterly wage) come from the Bureau of Labor Statistics Covered Employment and Wages program (ES-202).
Regression models were developed to determine the statistical relationship between the number of “core” benefits5 a firm offers and that firm’s level of employee turnover. Specifically, turnover is regressed6 on firm size, average quarterly wages, and the number of benefits offered for various industry groupings.7 Firm size and average quarterly wage are statistically controlled because it is likely that they also affect employee turnover. Holding firm size and average quarterly wage constant, we find a negative relationship between the number of benefits and turnover. In order to control for a firm’s industry and proportion of full-time employment, we developed separate models for each group (see bullet points for industry groupings).
Theoretically, we expect firms offering more core benefits to have lower turnover rates (net of other factors). In this analysis we developed a nine-point additive scale of benefits (0-8) ranging from no benefits to eight benefits. When offered, each of the core benefits adds an incremental digit to the benefits scale. This method assumes that all core benefits are equally important.8 Separate analyses were conducted on industry groupings for firms employing predominantly full-time employees and firms employing predominantly part-time employees.9 For predominantly full-time firms, the scale of benefits refers to benefits offered to the firm’s full-time employees. For predominantly part-time firms, the scale refers to benefits offered to the firm’s part-time employees. The following industry groupings are utilized in this article:
Figure 1 shows that benefits are significantly related to turnover among firms in goods-producing industries employing predominantly full-time workers. As hypothesized, turnover is reduced as more benefits are offered. On average, each additional benefit offered results in a .96 percentage point decrease (the regression coefficient for benefits) in turnover. Over the entire range of benefits, predicted turnover rates decrease from 22.8 percent turnover at zero benefits to 15.1 percent at eight benefits.
Figure 2 mirrors the same general pattern for services-producing industries. Note the substantially steeper regression slope for firms employing predominantly part-time employees in comparison to firms with predominantly full-time workers (-1.9 compared to -.6). Although both slopes are statistically significant, the reduction in turnover for each additional benefit offered is larger in firms employing predominantly part-time employees. According to these findings, benefits are a more powerful tool for reducing turnover among firms employing predominantly part-time workers.
On average, firms in both goods- and services-producing industries that offer more benefits have significantly less turnover. It also appears that, among firms in services-producing industries, those that employ predominantly part-time workers realize a sharper decline (i.e., a steeper negative regression slope) in turnover for each benefit offered than firms employing predominantly full-time workers.
R&P wondered whether these same patterns occur at lower levels of industry aggregation. Within services-producing industries, Retail Trade and lower-wage Services have among the highest rates of turnover.14 Further, both of these industries have experienced growth (particularly Services) in Wyoming’s economy relative to other industries over the last ten years.15 Finding ways to reduce turnover in these industries would help stabilize employment for many Wyoming workers. Thus, it is useful to determine if firms that offer more benefits in these sub-industry categories also have lower rates of turnover.
Figures 3 and 4 report predicted turnover rates for Retail Trade and lower-wage Services. Benefits significantly reduce turnover across all industry types shown in Figures 3 and 4. The effect of benefits on turnover in firms employing predominantly part-time workers in Retail Trade (-1.7) is only slightly higher (-1.4) than predominantly full-time firms (see Figure 3). The pattern for lower-wage Services shown in Figure 4 is similar to that of services-producing industries overall (see Figure 2, page 11), with firms employing predominantly part-time workers seeing a more pronounced decline in turnover for every additional benefit. Among the regression results presented in Figures 1, 2, 3 and 4, lower-wage Services firms employing predominantly part-time employees experience the largest percent decrease (-2.9) in turnover for every additional benefit offered.
Conclusions and Comments
Our analysis indicates that firms that offer more benefits have lower turnover for all industry groups examined.16 However, the effect of benefits on turnover varies somewhat by industry group. Adding benefits is a more effective way of lowering turnover among firms that have mainly part-time employees in comparison to those with mostly full-time employees. This may be the case because other positive factors such as the pace and nature of work, environmental conditions in the workplace, autonomy and greater levels of supervisory or general societal respect might mitigate the effect of benefits on turnover in predominantly full-time firms.
Even though providing benefits has been shown here to significantly reduce turnover, providing benefits or additional benefits may not be in the best economic interest of any given firm. To illustrate, a firm will not be economically motivated to provide benefits if turnover cost (i.e., cost of recruiting, hiring, training) is lower than benefit cost. It may be in the economic interest of government to encourage the provision of benefits among firms (e.g., tax incentives) if the cost of turnover to government (e.g., job training and placement services) is higher than the cost of facilitating the provision of benefits.
One question not addressed in this research is the relative importance of benefits on turnover compared to other factors (e.g., direct compensation). Additionally, the benefits survey is an ongoing process.17 In the future, sufficient data should exist to allow us to examine the relationship between benefits and turnover in greater industry detail.
1Communities may benefit from reduced turnover because employees with continuous employment may be less likely to need job training and other forms of governmental assistance. High levels of employee turnover in a community likely lead to higher residential mobility as unemployed individuals seek work. High residential mobility has been linked to a host of social ills including increased alcohol and drug usage, increased property and violent crime, and neighborhood disorder and decay.
2Wyoming Department of Employment, Research & Planning, Employee Benefits in Wyoming: 2000.
32000 Benefits Survey data are weighted to correct for over-sampling of large firms and non-response.
4Turnover, or exit, rates represent the average percentage of a firm’s workforce that exits quarterly. Rates are calculated across all known quarters for a given firm and theoretically can range from 0 to 100 percent. For further clarification on the methodology used to calculate turnover rates see, Tony Glover, “Turnover Analyses: Definitions, Process, and Quantification,” 2001.
5The scale of core benefits utilized here was delineated empirically using factor analysis and the eight items form a single additive scale (Alpha =.783 for the full-time benefits scale and Alpha = .848 for the part-time benefits scale). Other benefit types exist (maternity/paternity leave, disability insurance, etc.), however, they do not cluster together significantly with the “core” benefits and they have less predictive power for turnover.
6We use statistical models to explain and/or predict labor market phenomena. In this case, we want to explain and predict employee turnover. Thus, the rate of turnover is the dependent variable, meaning it is dependent upon other factors. The factors that we use to explain turnover (e.g., access to benefits) are called independent variables. In our model, access to benefits, firm size, and average quarterly wages are the inputs and an estimate of the rate of employee turnover is the output.
7The regression equation is mathematically defined as Ypred = a + b1 * (Firm Size) + b2 * (Average Quarterly Wage) + b3 * (Benefits). Firm Size and Average Quarterly Wage are set at the mean level within the groupings.
8Other metrics could be developed that weight each of the core benefits differently. However, for ease of interpretability and in the absence of theoretical or empirical data indicating the relative importance of the individual core benefits utilized here, R&P chose a simple additive scale.
9The method for determining the number of full- and part-time employees for firms in the 2000 Benefits Survey is outlined in Mark Harris and Krista Gerth, “Methods for Imputing Work Status,” 2003.
10Goods-producing firms include those in Agriculture, Mining, Construction, and Manufacturing.
11There are too few firms employing predominantly part-time workers in goods-producing industries to conduct a reliable regression analysis.
12Lower-wage Services firms, two-digit Standard Industrial Classification codes (SIC), include Business Services (73), Museums (84), Social Services (83), Amusements & Recreation Services (79), Membership Organizations (86), Personal Services (72), Hotels & Other Lodging Places (70), and Motion Pictures (78). These two-digit SICs have average weekly wages (based on 2000 published data) lower than $400 dollars (i.e., below the mean level for all two-digit Services major groups). For more information, see <http://doe.state.wy.us/LMI/00202pub/00t28.htm>.
13Services-producing firms include those in Transportation, Communications, & Public Utilities (TCPU); Wholesale Trade; Retail Trade; Finance, Insurance, & Real Estate (FIRE); Services; and Government.
14Wyoming Department of Employment, Research & Planning, Market Dynamics from Administrative Records, December 2002.
15Mark Harris, “Is Wyoming’s Economy Diversifying and Is Economic Diversity in Wyoming Desirable,” Wyoming Labor Force Trends, September 2002, pp. 1-9.
16Our regression models simultaneously account for industry, the proportion of full- and part-time employees, firm size, and firms average quarterly wage when ascertaining the effect of benefits availability on turnover. However, the reader should be aware that, although our regression analysis shows support for the theoretical contention that offering benefits reduces turnover, it is possible that other unmeasured factors that we have not included may account for the effect of benefits on turnover. In other words, benefits may be serving as a “proxy” for other important unmeasured characteristics and not actually be significantly predictive of employee turnover. To illustrate with a facetious example, let us assume that firms offering benefits have bosses with positive attitudes and firms not offering benefits have bosses with negative attitudes. Let us also assume that boss “attitude” is what primarily reduces turnover. In such a case, if we measure the effect of boss attitude and benefits availability on turnover simultaneously, benefits availability will not be shown to significantly reduce turnover but boss attitude will. Because of the high correlation between boss attitude and benefits availability, failure to account for boss attitude allows benefits availability to proxy for boss attitude and will lead to the erroneous conclusion that benefits availability significantly reduces turnover when in fact it does not.
17Wyoming Department of Employment, Research & Planning, Employee Benefits in Wyoming: 2001.
Table of Contents | Labor Market Information | Wyoming Job Network | Send Us Mail
designed by Julie Barnish.
Last modified on by Susan Murray.