The Wage and Job Impacts of Hospitals on Local Labor Markets

Abstract

This study examines the impact of hospitals on local labor markets in rural and urban counties. We measure the ability of hospitals, particularly in rural communities, to attract nonhealth-related employment and provide higher wage jobs to residents based on their education level. Results find hospital employees with an associate’s degree can expect a 21.4% wage premium, when compared with alternative opportunities, and those with a bachelor’s degree can earn 12.2% more working in a hospital. Hospitals are shown to be positively related to overall employment as well as exhibit positive employment spillover. For rural counties, a short-term general hospital is associated with 559 jobs in the county, 60 of which are hospital based and 499 are non–health care related. With the positive benefits on wages and non–health care job growth, hospitals have measurable positive labor market outcomes above their primary objective of providing health care access, particularly in rural counties.

Keywords

regional sciences rural development job creation

In 2011, the Congressional Budget Office predicted that hospitals that qualify for certain federal subsidies will cost the federal government $23 billion over the 2012 to 2016 period.¹ Under normal market conditions, failing businesses are an indicator that the market does not have a sufficient preference for the good or service, given the market price. However, because of moral hazard, adverse selection, and price distortion, hospitals do not function under normal market conditions. Thus, national and local governments have been willing to allocate resources to hospitals that otherwise would be losing money under the justification that hospitals provide numerous positive benefits for the local community.

On a national level, hospitals are an important part of the economy. In 2013, hospitals produced $1.2 trillion in value add and $2.0 trillion in total economic output.² On a local level, hospitals provide a myriad of benefits that range from their primary function of providing medical access for residents, acting as an “export-base industry” through the inflow of federal Medicare and Medicaid payments, and supplying high-skill high-wage employment (Nelson, 2009). Hospitals may also serve as an export base industry if they attract patients and visitors from outside the county or region either through delivering higher quality or a broader array of services than other nearby locations.³ While the general qualitative benefits of hospitals are intuitively apparent, estimation of these impacts involves some effort to isolate each one.

This study examines the impact of hospitals on local communities by measuring the relationship between hospitals and local labor markets in rural and urban areas. When considering local labor markets, by far two of the most important factors are wages of local residents and employment opportunities in the community. Our formal research objective is to measure if hospitals, particularly in rural communities, attract nonhealth-related employment, and if hospitals provide higher wage jobs to residents based on their education level.

Previous studies have primarily measured the economic impact of hospitals with input–output analysis. Because of the limitations, however, particularly for rural counties (Holmes, Slifkin, Randolph, & Poley, 2006), we measure employment and wage outcomes due to the presence of a hospital using two alternative model specifications. First, the differences in wages among health care versus non–health care workers is measured by statistically modeling workers’ earnings conditional on educational attainment for years 2000 and 2010. This approach is necessary because input–output models use fixed prices and cannot, by design, be used to estimate wage premia. Second, the relationship between hospitals and nonhealth-related employment for years 2001 to 2010 for urban and rural counties is measured using regression analyses.

The remainder of this study is organized as follows: The second section is a review of selected previous work, the third describes the data and their sources, and the fourth explains the empirical methodology and results. A conclusion and summary close the study. Because we are running separate analyses for employment and wages outcomes, the third and fourth sections have separate subsections for the employment and wage analyses.

Background and Previous Studies

The economic impacts of hospitals in urban versus rural communities are not uniform. While urban communities usually have multiple hospital options within a city that can cater to differing population and health demographics, rural counties generally rely on a single hospital provider. This difference is particularly evident when a rural hospital closure is considered. When there are rumors of an urban hospital closure, the conversation is generally focused on health access for the population currently utilizing that hospital’s services. In contrast, hospital closures in rural communities seem to include an additional serious concern for the closure’s impact on the future of local economic development. The typical fear is that employers will not open new facilities in counties without a hospital (Dillard, 2015).

One example of this is in Clayton County, Georgia: When the area’s only hospital was in jeopardy of being closed, community leaders fought together to keep it open. The mayor worried that “Our citizens need to have some health care and somewhere to go” and noted that losing one of the county’s largest employers would “create a devastating domino-effect” (Joyner, 2014).

Generally, low population counties have been second to their urban counterparts in terms of job-growth-linked income gains (McGranahan & Beale, 2002). Although rural counties have the benefits of a lower cost of living and inexpensive land, it is difficult to compete with larger cities in terms of business support amenities, agglomeration, and access to larger pools of high-skilled labor. One major aid in rural economic development is the health sector. Generally, health jobs are second only to the education sector in terms of total employment for rural counties. Additionally, the health sector does not provide average jobs, but high-wage high-skilled jobs. This helps build a strong tax base and stimulates other local businesses when those wages are spent locally (Doeksen, Cordes, & Shaffer, 1992).

While some studies have failed to find a relationship between hospital closure and economic outcomes (Pearson & Che, 2003; Probst, Samuels, Hussey, Berry, & Ricketts, 1999; Stensland, Moscovice, & Christianson, 2002), there seems to be a consensus among rural policy makers that hospitals are critical for a rural county’s economic vitality. Typically, three additional benefits are attributed to hospitals in rural communities beyond access to care: hospitals bring in high-skilled, high-paying jobs; hospitals are an important amenity to potential migrants; and hospitals are critical for attracting future business growth (Christianson & Faulkner, 1981; Cordes, Sluis, Lamphear, & Hoffman, 1999; Doeksen et al., 1992; Doeksen, Johnson, & Willoughby, 1997; Doeksen, Loewen, & Strawn, 1990; Doeksen & Schott, 2003; Mick & Morlock, 1990).

Hospitals could also help local employees through interindustry wage differentials. Numerous studies indicate that wages persistently vary across industries and businesses for workers with the similar characteristics (e.g., Dickens & Katz, 1987; Groshen, 1991; Krueger & Summers, 1988; Thaler, 1990). Thus, if a hospital that hires a large percentage of a rural population compensates its employees well, it could influence local labor market norms about wages and other employer practices. It may be that other businesses would have to adopt such practices if they wished to acquire high skilled labor.

Whereas these studies are among an extensive literature measuring the economic impact of hospitals, previous research has predominantly relied on input–output analysis. An exception to this is Lindrooth, Lo Sasso, and Bazzoli (2003), who analyze urban hospital closures using regression analysis. They find that when alternative care is available, there are potential efficiency gains when struggling urban hospitals close. Holmes et al. (2006) measure the effect of rural hospital closures on rural communities using a fixed-effect regression analysis. They find that hospital closures in communities with only one hospital lead to higher unemployment and lower per capita income. Bartik and Erickcek (2008) examine the relationship between the health sector and economic activity within a metropolitan area and find above-average wages for health sector employees, holding worker characteristics constant. Capps, Dranove, and Lindrooth (2010) measure residents’ welfare after a hospital closure and find total surplus (a measure of aggregate social welfare) in the local community can decline following hospital closures. Brooks and Whitacre (2011) similarly used regression analysis and found that having a critical access hospital in a community leads to higher levels of retail activity.

Our study compliments previous work by measuring the economic impact of hospitals on local labor markets in a broader sense than possible with input–output models. In particular, we aim to capture nonproduction channel impacts such as how the amenity effects of having a hospital in a community might attract retiree migrants and attract new businesses (Mandich & Dorfman, 2016).

This study has two model specifications, one to measure the influence of hospitals on wage premiums and another for employment spillovers. For the wage analysis, we choose a linear regression model and analyze individual-level wages for both years 2000 and 2010. Measuring wages in these two time periods allows us to examine any changes in hospital impact on wages over time. This is particularly interesting, given the recent significant recession from 2007 to 2009. Because of these individual-level wage data, we are able to interact education with hospital employment to measure the impact of health care employment on wages for various skill levels. This interaction has not been thoroughly explored previously, as most of the employment conversation around hospitals has centered on the ability of hospitals to provide high-skilled, high-wage employment such as doctors. This study contributes to the literature by specifically exploring hospital wage impacts on low-skilled workers. We additionally contribute to the literature by measuring hospital employment spillover effects using regression analysis.

Employment outcomes are examined at the county level using panel data for 2002 to 2012. To control for local unobserved characteristics, regression analysis with county-level fixed effects was the chosen model specification. By running two regressions, one for total employment and another for local health care sector employment, we are able to calculate the total non–health care employment that is attributed to having a hospital in the county. This is the spillover effect. For clarity, this spillover is not equivalent to a job multiplier; rather, job spillover can be thought of as the long-run consequence of businesses being attracted to a location by the amenity of having a hospital. Finding a positive and significant employment spillover effect from hospitals would confirm rural policy makers’ intuition that hospitals attract employment that is non–health care related, and that hospitals are an important factor for rural communities’ economic development prospects.

Data

Because we have two separate analyses, one at the individual level and one at the county level, the data are from two separate sources. As described further, individual-level Integrated Public Use Microdata Series (IPUMS) data are used for wage premium calculations, while county-level hospital employment from the U.S. Bureau of Economic Analysis (BEA) is used for measuring employment spillover.

Wages

Individual-level data used in the wage analysis include the 5% sample of the 2000 census and the 1% sample of the 2010 American Community Survey (ACS) from IPUMS. Summary statistics are reported in Table 1. To compute the dependent variable of log wage, the log of respondent’s annual pretax wage and salary income was divided by the average hours and weeks worked past year. Because these are self-reported and not administrative data, there is naturally noise in the data. We thus exclude all values below $5 per hour and then top code obvious outliers, conditional on education level, after graphing income by education level. This is not a perfect measurement of hourly income. Because we use the full 5% IPUMS sample that consists of more than 6 million observations in 2000 and more than 1.8 million observations from 2010, we rely on the law of large numbers that, although some estimates are imperfect, the averages of the final data series should be close to the true value.

Table 1.

Wage Descriptive Statistics.

	2000				2010
	M	SD	Min	Max	M	SD	Min	Max
Log wage	2.642	0.590	1.61	5.298	2.881	0.651	1.609	9.462
Married	0.601	0.490	0	1	0.557	0.497	0	1
Black	0.099	0.299	0	1	0.097	0.296	0	1
Female	0.471	0.499	0	1	0.500	0.500	0	1
Age	39.1	11.8	18	64	41.0	12.9	18	64
Age²	1671.9	946.3	324	4,096	1903.5	1070.6	324	4,096
Urban	0.300	0.456	0	1	0.327	0.333	0	1
Less than high school	0.127	0.332	0	1	0.073	0.259	0	1
High school	0.279	0.449	0	1	0.223	0.416	0	1
Some college	0.253	0.435	0	1	0.298	0.457	0	1
Associate’s	0.076	0.265	0	1	0.082	0.274	0	1
Bachelor’s	0.173	0.378	0	1	0.195	0.397	0	1
Postgraduate	0.092	0.289	0	1	0.130	0.336	0	1
Hospital employee	0.044	0.204	0	1	0.046	0.210	0	1

The independent variables were chosen based on individual-level characteristics shown to affect earnings that include race, age, sex, education, and working in an urban area (defined here as in a city or incorporated area with a population of more than 2,500). Because we want to measure whether there is a potential for a wage premium for historically underpaid workers based on race, we include a variable to distinguish Black workers in our final estimation. Also, we only include workers of the traditional working age of above 18 and less than 65 years.⁴ Because education levels strongly predict future earnings (Card & Lemieux, 2000), education is categorized into less than a high school education, having a high school education or equivalent, some college, an associate’s degree, a bachelor’s degree, and a postgraduate degree. Defining education into such categories allows us to estimate how employment in the health sector affects people at various skill levels and not just highly skilled doctors and surgeons. We define hospital workers according to the North American Industrial Classification System (NAICS) code of 622. Finally, because wages reflect cost of living, we include a dummy variable, urban, for whether the respondent works in an urban city center, based on the IPUMS classification.

We admit that this analysis is not perfect because people with the same education levels in different industries may have different mixes of majors and job skills, leading to some wage differentials. We believe, however, that hospitals employ a wide enough array of people (e.g., health professionals, accountants, supply chain managers, custodians, receptionists) for a meaningful comparison.

Employment

County-level data for the employment analysis come from the U.S. Department of Health and Human Resources Area Resource File and the BEA. Summary statistics are listed in Table 2. The Area Resource File is a collection of data from more than 50 sources such as the American Hospital Association, Bureau of Labor Statistics, and National Center for Health Statistics and contains many county health and population characteristics. All county-level characteristics in the analysis thought to affect employment were collected from this source.

Table 2.

Summary Statistics of Employment by County Type, 2002-2010.

		Count	M	SD	Min	Max
Urban	Total employment	6,426	71,230	88,566	519	620,335
	Health care employment	4,871	9,751	10,839	12	75,854
	Long-term hospital	6,426	0.15	0.44	0	4
	Short-term nongeneral hospital	6,426	0.61	1.30	0	13
	Short-term general hospital	6,426	1.82	1.76	0	12
	Log income	6,426	10.31	0.20	9.56	11.29
	Unemployment	6,426	5.68	2.40	0	28.20
	% Health care employment	4,871	0.10	0.04	0	0.37
Rural	Total employment	17,843	12,822	22,196	60	845,120
	Health care employment	9,567	1,472	2,447	0.00	62,986
	Long-term hospital	17,613	0.04	0.19	0	2
	Short-term nongeneral hospital	17,613	0.04	0.21	0	3
	Short-term general hospital	17,613	0.98	0.72	0	6
	Log income	17,843	10.22	0.22	9.30	11.73
	Unemployment	17,613	6.13	2.88	0.60	25.50
	% Health care employment	9,567	0.09	0.04	0	0.29

Total full-time and part-time employment data by NAICS industry were obtained from the BEA. Our employment analysis uses panel county-level data for 2002 to 2010 and is classified into aggregate employment and health care employment, defined by Standard Industry Classification code 8060. To more effectively capture the differences between urban and rural counties, we run separate analyses based on the population of the county. Using the 2013 Urban Rural Continuum Code, we run two analyses—one for urban counties (code = 2, 3) and a second for rural counties (code ≥ 4). Because of confidentially concerns, some counties have suppressed values for health care employment; thus, we are missing those data. Although we would prefer to have data on all counties, the excluded counties are likely the smallest ones, where we would expect the largest employment benefits from hospitals. Because the literature has shown that health care employment has positive benefits for rural counties, our results are likely a lower bound, with the actual effect as large or larger than we estimate.⁵ Note that because hospital closings are not a policy concern in large cities and difficulties in scale factors such as hospitals ranging from small to extremely large, we do not include any analysis of the most urban counties (code = 1).

While most of the variables in Table 2 are self-explanatory, a few deserve further explanation. To avoid a simultaneity bias, all hospital variables are lagged by 10 years. Thus, it can be assumed that a business moved to an area with full knowledge of the health access and not the reverse. Because a major research hospital and a critical access hospital have different economic implications, we include three types of hospital classifications in the analysis: short-term general hospitals, short-term nongeneral hospitals, and long-term hospitals. A short-term general hospital can be defined as having facilities and staff to provide diagnosis, care, and treatment of a wide range of acute conditions, whereas short-term nongeneral hospitals provide treatment for a limited special group of acute conditions. Long-term hospitals have the infrastructure and personnel for the diagnosis, care, and treatment of a wide range of chronic diseases and have an average inpatient length of stay longer than 25 days. The remaining independent variables include the log of median income in the county, the unemployment rate in the previous year, and the percentage of people employed in health care. The percentage of health care employment in the county is included to identify counties that particularly rely on hospitals as a major driver of employment opportunities.

Empirics

Wages

Regression analysis is used to measure the individual-level wage premium associated with working in a hospital. A simple model specification uses an individual’s log wage as the dependent variable, with worker-level characteristics as independent variables. Specifically, we mathematically represent this as

y = β_{0} + β_{1} X + β_{2} h o s p + u

where y = log wage, $X$ is a vector of individual-level characteristics, $h o s p$ is a dummy for being employed within a hospital, and u is a robust standard error. The results from this simple model are reported in Table 3, column 1, and are consistent with the established labor literature with women earning less than men, and higher educational attainment producing a higher hourly wage. For this study, the analysis of greatest interest is how much more a worker can expect to earn if employed in a hospital conditional on one’s educational attainment. As seen in Model 1 for 2000 and 2010, overall employment in the health industry induces a 7.4% and a 16.9% hourly wage premium over workers in other sectors. Essentially, in 2010 if Person A is earning the national average of $51,000 in an alternative industry, by switching to work in a hospital, Person A could expect to earn $59,619 or a wage premium of $8,619.

Table 3.

Hospital Employment Wage Premium.

Dependent variable: Log wage	2000		2010
Dependent variable: Log wage	(1)	(2)	(1)	(2)
Married	0.075*** (152.74)	0.075*** (152)	0.106*** (84.72)	0.106*** (84.44)
Black	−0.031*** (−40.99)	−0.029*** (−37.38)	−0.063*** (−33.33)	−0.059*** (−30.51)
Female	−0.221*** (−476.79)	−0.225*** (−479.04)	−0.203*** (−176.56)	−0.207*** (−177.18)
Age	0.045*** (339.32)	0.045*** (339.38)	0.052*** (157.08)	0.052*** (157.25)
Age²	−0.0004*** (−259.47)	−0.0004*** (−259.39)	−0.0005*** (−126.14)	−0.0005*** (−126.18)
Urban	0.086*** (126.29)	0.086*** (126.78)	0.084*** (51.63)	0.084*** (51.9)
High school	0.128*** (181.6)	0.130*** (182.75)	0.172*** (85.23)	0.175*** (85.68)
Some college	0.278*** (368.29)	0.279*** (365.06)	0.349*** (163.9)	0.352*** (163.23)
Associate’s	0.326*** (338.37)	0.310*** (311.04)	0.398*** (160.84)	0.381*** (149.45)
Bachelor’s	0.574*** (674.69)	0.572*** (658.5)	0.666*** (301.75)	0.662*** (294.96)
Postgraduate	0.773*** (701.01)	0.774*** (685.59)	0.931*** (359.37)	0.931*** (352.78)
Hospital employee	0.074*** (68.03)	−0.048*** (−10.14)	0.169*** (66.1)	0.015 (1.05)
High school * Hospital employment		−0.065*** (−13.71)		−0.054*** (−3.65)
Some college * Hospital employment		−0.014** (−2.90)		−0.020 (−1.36)
Associate’s * Hospital employment		0.170*** (35.27)		0.214*** (14.59)
Bachelor’s * Hospital employment		0.063*** (13.15)		0.122*** (8.42)
Postgraduate * Hospital employment		0.026*** (4.56)		0.077*** (4.85)
Female * Hospital employment		0.119*** (43.48)		0.113*** (17.97)
Black * Hospital employment		−0.003 (−0.88)		−0.025*** (−3.40)
Constant	1.344*** (546.89)	1.345*** (547.47)	1.269*** (196.35)	1.270*** (196.43)
R ²	.297	.299	.331	.332
N	5,918,257	5,918,257	1,250,703	1,250,703

Note. t Statistics in parentheses.

p < .05. **p < .01. ***p < .001.

However, we should be initially cautious of this result. It could be that this wage premium is due to the disproportionate amount of high earners the health industry employs (i.e., surgeons) and may not be applicable to the total population. To test for this, we also model whether one receives a wage premium for working in the health industry based on one’s level of education. By interacting hospital employment and education level, we can compare for each level of education how much more a person can earn being employed in a hospital. This allows us to analytically compare two people with a specific level of education, except one works in the health industry and the other does not, and measure who makes more per hour and by how much. This can be done using a slightly modified regression model:

y = β_{0} + β_{1} X + β_{2} h o s p + β_{3} h o s p * e d u c + u .

Education-specific results are shown in Table 3, column 2. We find a 21.4% premium in 2010 for hospital employees with an associate’s degree compared with other people with an associate’s degree. Similarly, a person with a bachelor’s degree would earn 12.2% more, and one with postgraduate education can expect a 7.7% wage premium over others with the same level of education. All three of these results are statistically and economically significant. In monetary terms, if a person with an associate’s degree was annually making $30,000 in an average-earning industry, they could make $36,420 working in a hospital. In this scenario, hospital employment would have a $6,420 wage premium for those with an associate’s degree. Considering that associate’s degrees are generally 2-year programs with flexible program designs, this has major policy implications for creating medium-skilled high-paying employment.

We find mixed results in hospital wage outcomes based on individual characteristics. When measuring the presence of a hospital wage premium among Black workers, there is no significant relationship in 2000 and in 2010, there is a small negative relationship. We do, however, find an 11.3% hospital wage premium among women in 2010. Hospitals have a strong demand for historically female-dominated jobs such as administrative positions and nursing. Thus, hospitals have not only been employing a large proportion of females historically but are also currently providing women with better paying jobs relative to outside options with the same education level.

Another interesting finding appears when comparing the 2000 and 2010 results. While these results are not perfectly comparable, as the 2000 results are based on census long-form data and 2010 relies on the ACS,⁶ we see that none of the signs change. Incidentally, all the hospital employment magnitudes increase in 2010 compared with 2000. This is one signal for the durability of a hospital’s potential for positive economic impact. We also see a stronger link between education level and wage outcomes over time. For 2010, someone with a bachelor’s degree can expect to earn 66.6% more per hour than someone without a high school diploma compared with 57.4% in 2000. This implies that the wage premium for higher education has increased in the past 10 years.

These results have important implications for policy makers. Our findings show that hospitals not only provide high-income high-skilled employment (e.g., surgeons and doctors) but also have positive impacts for those with lower levels of education. For communities with a hospital present, workers with 2 or fewer years of post–high school education can find employment that pays significantly more than other opportunities for those with an associate’s degree. Thus, hospital workers with an associate’s, bachelor’s, or postgraduate degree can expect a positive wage premium compared with outside opportunities.

Employment

Our next level of analysis measures the association between hospitals and jobs. We begin the analysis with Figure 1, which depicts the percentage change in national total employment and in national health sector employment from 2002 to 2011. Two things are particularly striking. First, although health employment growth slows in 2004 and 2009 to 2011, hospital employment never drops. Over this time period, which includes a severe recession, health care employment is continuously increasing. In contrast, national employment does drop during 2008 to 2010. This would suggest that, in the aggregate, being employed in health care helps shield one from negative economic shocks, such as a recession. This is true for both urban and rural health sectors, as shown in Figure 2. Essentially, hospital jobs are more “recession proof” than alternative employment opportunities.

Figure 1.

Percentage change in total and hospital employment, 2002-2011.

Figure 2.

National hospital employment change by rural and urban.

To measure the implications of hospitals on local job markets, we measure county employment and the presence of a hospital for 2001 to 2011. We use a linear regression model with panel clustered standard errors, time dummies, and county fixed effects. Mathematically, this is represented as:

Y_{i t} = X_{i t} β + δ y r_{t} + f_{i} + u_{i t}

where $Y_{i t}$ is the number of employed 18- to 64-year-old individuals in the county, and $X_{i t}$ contains annual county-level characteristics, including type and number of hospitals present. Because there was a substantial recession during the selected time period, annual dummies, $y r_{t}$ , are included to control for annual shifts in local employment. Similarly, because there will naturally be unobserved heterogeneity given the nature of county-level data, county fixed effects, $f_{i}$ are also included, as well as county clustered standard errors $u_{i t}$ .⁷

Another way we decrease the unobserved heterogeneity is by categorizing the data based on how urban the county is. As seen in Table 2, urban counties have an average of 1.82 short-term general hospitals, whereas rural counties have an average of only 0.98 short-term general hospitals. By estimating Equation (3) separately for urban and rural counties, we can observe how hospitals are affecting employment, particularly in rural counties where there is likely only one hospital.

As shown in Table 4, county characteristics assumed to affect employment perform generally as expected. For all counties, the unemployment rate in the previous year is negatively associated with employment. Similarly, we also see that, compared with the omitted year of 2001, the following years have more jobs than 2001 with the exception of the postrecession years of 2008 and later for urban and rural counties where the magnitude is noticeably smaller than the previous trend. As expected, log of income was universally positive for total employment in all counties; however, income was actually negative in both urban and rural counties. One explanation for this is that in areas outside major cities it is possible that lower incomes correlate with higher demand for health services because of provision of Medicaid (and perhaps also Medicare).

Table 4.

Total and Hospital Employment by County Type, 2002-2010.

	Urban		Rural
	Total	Hospital employment	Total	Hospital employment
Long-term hospital	2,451*** (3.81)	292** (2.3)	548*** (2.46)	53 (1.6)
Short-term nongeneral hospital	1,116*** (3.67)	264*** (3.75)	312** (1.98)	95*** (2.84)
Short-term general hospital	1,045* (1.67)	66 (0.5)	559*** (3.83)	60*** (2.67)
Log income	19,892*** (5.29)	−1,769*** (−2.92)	1,997*** (7.5)	−103*** (−3.21)
Unemployment	−651*** (−4.02)	−77** (−2.07)	−99*** (−8.99)	−6*** (−3.07)
% Health care employment	−53,395* (−1.69)	56,239*** (6.53)	−12,224*** (−5.98)	9,858*** (12.27)
Year 2003	925*** (4.23)	297*** (8.32)	49*** (2.36)	28*** (8.41)
Year 2004	2,024*** (5.31)	586*** (9.17)	179*** (5.33)	52*** (9.93)
Year 2005	2,648*** (5.09)	793*** (8.84)	284*** (6.1)	73*** (10.66)
Year 2006	3,506*** (4.92)	1,123*** (9.01)	424*** (7.32)	104*** (12.16)
Year 2007	4,153*** (4.55)	1,474*** (8.86)	459*** (6.16)	141*** (12.42)
Year 2008	2,866*** (2.56)	1,656*** (7.89)	230*** (2.5)	154*** (11.13)
Year 2009	1,798* (1.73)	1,574*** (8)	−53 (−0.63)	139*** (11.01)
Year 2010	3,039** (2.33)	1,916*** (8.54)	120 (1.23)	167*** (10.87)
Constant	−112,302*** (−2.88)	21,411*** (3.29)	−5,319 (−1.99)	1,465*** (4.44)
N	4,871	4,871	9,427	9,427
R ²	.3521	.4736	.2263	.4386

Note. t Statistics in parentheses.

p < .05. **p < .01. ***p < .001.

Hospitals are positively associated with employment among all counties. Rural counties with short-term general hospitals can attribute 559 jobs, while urban counties can attribute 1,045 jobs. Although these results seem intuitive given that hospitals are a major source of employment, an aspect of hospital economic impact we specifically want to measure is the number of non–health care jobs that can be attributed to the presence of a hospital.

We measure the non–health care employment gains related to hospitals applying the following methodology. First, we save the results from Equation (3) of the impact of hospitals on county employment. Then, we change the dependent variable $Y_{i t}$ in Equation (3) to county health care employment. This gives us the number of health-related jobs in the county because a hospital is present in the county. We can then subtract the impact hospitals have on health-specific employment from total employment. This difference is the net nonhealth-related job gains associated with having a hospital in the county. For example, as seen in Table 5, a short-term general hospital is associated with 559 total jobs in an average rural county. The table also shows that the expected gain in health care–only employment associated with having a hospital in a rural county is 60 jobs for a short-term general hospital. Thus, the number of expected non–health care jobs a rural county would gain from a short-term general hospital is 499 jobs. Similarly, a short-term nongeneral hospital is expected to produce a gain of 216 nonhealth-related jobs for a rural county.

Table 5.

Total and Hospital Employment by County Type, 2002-2010.

	Urban			Rural
	Total	Hospital employment	Difference	Total	Hospital employment	Difference
Long-term hospital	2,451	292	2,160	548	53	495
Short-term nongeneral hospital	1,116	264	852	312	95	216
Short-term general hospital	1,045	66	979	559	60	499

Hospital job spillover is similarly present in the urban counties. Specialty hospitals had particularly high job gains. Short-term nongeneral hospitals were responsible for 264 hospital jobs in urban counties and 852 nonhospital-related jobs. Urban counties were also found to benefit from having long-term-care hospitals, which were associated with 292 hospital jobs in the county and 2,160 nonhospital-related jobs. Whereas urban hospitals do not usually employ the same percentage of the population as in rural counties, we do find that hospitals are positively contributing to an urban county’s local job market beyond the direct employment within the hospital.

To place these results in context, we can compare the spillover job gains with the average total employment in each category of the county. Rural counties have an average of 11,895 jobs and urban counties have an average of 70,983 jobs. Using the spillover gains from short-term general hospitals, these results mean that one additional short-term general hospital can be expected to raise non–health care employment in counties by 4.2% for rural counties and by 1.4% for urban counties. These gains are definitely nontrivial and show why local governments, especially in rural areas, are eager to keep their local hospitals open.

These results have definite policy implications for numerous reasons. First, these results show that businesses are attracted to communities that have access to health services, thus keeping a hospital open has consequences not only for the health access of residents but also has consequences on the health of the local labor market and economy. This should inform policy makers being asked for subsidies to keep rural hospitals open. Second, looking at the relationship between hospitals and nonhealth employment gives a much clearer picture of the true economic spillover effect of hospitals. For example, if having a hospital in County X is associated with 60 jobs but the hospital employs 60 people, this hospital has no spillover effect because all of its job creation is internal. As Table 5 shows, however, this is not the case, as all counties have positive spillover employment gains from having a hospital, meaning they increase employment more than just their direct employment in the health care sector. Beyond traditional multiplier effects on hospital employee spending, hospitals appear to serve as an amenity in the business attraction process.

Conclusions

With positive benefits on wages and nonhealth-related jobs growth, hospitals have measurable positive economic outcomes above their primary objective of providing health care. Our analysis finds that hospitals provide high wage jobs not only for the most educated population but also among those with 2- and 4-year degrees. In 2010, hospital employees with an associate’s degree could expect a 21.4% wage premium compared with those in other industries with the same level of education. Similarly, a person with a bachelor’s degree would earn 12.2% more working in a hospital compared with outside opportunities.

In terms of jobs, overall employment is positively related to a strong health care presence in rural and urban counties. For rural counties, a short-term general hospital is associated with 559 jobs in the county, 60 of which are in health care and 499 that are non–health care related. Urban counties were also found to benefit from having long-term-care hospitals, which were associated with 292 hospital jobs in the county and 2,160 nonhospital-related jobs. The strong positive spillover effect that hospitals have on non–health care employment suggests that hospitals are an important institution for job creation. Thus, hospital closures would not only affect direct health care employment but also many other jobs in the community.

At a time when smart job creation and growth is critical, it appears the health sector can play a vital role, particularly for the future economic growth of rural counties. In short, hospitals have significant positive economic impacts on their local communities as measured through these labor market outcomes. It may be particularly advantageous for local policy planners to consider these outcomes when pursuing new businesses and striving to build strong communities. These results also have implications for local governments wrestling with decisions about whether to provide subsidies to keep their local hospitals open. When asked to provide local governmental funding to keep a hospital operating, these results help give local policy makers the evidence they need to make informed decisions.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Anne M. Mandich is an analyst at Thumbtack, and has a PhD in agricultural and applied economics from the University of Georgia.

Jeffrey H. Dorfman is a professor of agricultural and applied economics from the University of Georgia, and a frequent consultant to local governments and developers on land use issues and the impact of land uses on local government budgets.

References

Bartik

T. J.

Erickcek

G. A.

(2008). “Eds & meds” and metropolitan economic development. Employment Research Newsletter, 15(1), 1. Retrieved from http://research.upjohn.org/cgi/viewcontent.cgi?article=1085&context=empl_research

Brooks

Whitacre

B. E.

(2011). Critical access hospitals and retail activity: An empirical analysis in Oklahoma. Journal of Rural Health, 27, 29-38.

Capps

Dranove

Lindrooth

R. C.

(2010). Hospital closure and economic efficiency. Journal of Health Economics, 29, 87-109.

Card

Lemieux

(2000). Can falling supply explain the rising return to college for younger men? A cohort-based analysis (No. w7655). Retrieved from http://davidcard.berkeley.edu/papers/men-return-college.pdf

Christianson

J. B.

Faulkner

(1981). The contribution of rural hospitals to local economies. Inquiry, 18, 46-60.

Cordes

Sluis

Lamphear

Hoffman

(1999). Rural health research: Rural hospitals and the local economy: A needed extension and refinement of existing empirical research. Journal of Rural Health, 15, 189-201.

Dickens

Katz

L. F.

(1987). Inter-industry wage differences and theories of wage determination (NBER Working Paper No. w2271). Retrieved from https://ssrn.com/abstract=227501

Dillard

(2015, February 14). Dying rural hospitals affect most vulnerable. Tyler Morning Telegraph. Retrieved from http://www.tylerpaper.com/TP-News+Local/213794/dying-rural-hospitals-affect-most-vulnerable

Doeksen

G. A.

Cordes

Shaffer

(1992). Health care’s contribution to rural economic development. Unpublished paper prepared for Federal Office of Rural Health.

10.

Doeksen

G. A.

Johnson

T. G.

Willoughby

(1997). Measuring the economic importance of the health sector on a local economy: A brief literature review and procedures to measure local impacts. Starkville, MS: Southern Rural Development Center.

11.

Doeksen

G. A.

Loewen

R. A.

Strawn

D. A.

(1990). A rural hospital’s impact on a community’s economic health. Journal of Rural Health, 6, 53-64.

12.

Doeksen

G. A.

Schott

(2003). Economic importance of the health-care sector in a rural economy. Rural and Remote Health, 3(1), 135.

13.

Gage

(2006). Comparison of census 2000 and American Community Survey 1999–2001 estimates: San Francisco and Tulare counties, California. Population Research and Policy Review, 25, 243-256.

14.

Groshen

E. L.

(1991). Sources of intra-industry wage dispersion: How much do employers matter? Quarterly Journal of Economics, 106, 869-884.

15.

Holmes

G. M.

Slifkin

R. T.

Randolph

R. K.

Poley

(2006). The effect of rural hospital closures on community economic health. Health Services Research, 41, 467-485.

16.

Joyner

(2014, March 4). Clayton OKs tax package to aid hospital. The Atlanta Journal-Constitution. Retrieved from http://www.ajc.com/news/local/clayton-oks-tax-package-aid-hospital/LBNq6clljpKRTrtKWf2M9L/

17.

Krueger

A. B.

Summers

L. H.

(1988). Efficiency wages and the inter-industry wage structure. Econometrica, 56, 259-293.

18.

Lindrooth

R. C.

Lo Sasso

A. T.

Bazzoli

G. J.

(2003). The effect of urban hospital closure on markets. Journal of Health Economics, 22, 691-712.

19.

Mandich

A. M.

Dorfman

J. H.

(2016). Elderly migration: Spatial considerations of amenity and health access drivers. Journal of Regional Science, 56, 96-133.

20.

McGranahan

D. A.

Beale

C. L.

(2002). Understanding rural population loss. Rural America, 17(4), 2-11.

21.

Mick

S. S.

Morlock

L. L.

(1990). America’s rural hospitals: A selective review of 1980s research. Journal of Rural Health, 6, 437-466.

22.

Nelson

(2009). Are hospitals an export industry? Empirical evidence from five lagging regions. Economic Development Quarterly, 23, 242-253.

23.

Pearson

D. R.

Che

(2003). The impact of rural hospital closure on the economic health of the local communities. Texas Journal of Rural Health, 21(3), 46-51.

24.

Probst

J. C.

Samuels

M. E.

Hussey

J. R.

Berry

D. E.

Ricketts

T. C.

(1999). Economic impact of hospital closure on small rural counties, 1984 to 1988: Demonstration of a comparative analysis approach. Journal of Rural Health, 15, 375-390.

25.

Stensland

Moscovice

Christianson

(2002). Future financial viability of rural hospitals. Healthcare Financing Review, 23(4), 175-188.

26.

Thaler

R. H.

(1990). Anomalies: Saving, fungibility, and mental accounts. Journal of Economic Perspectives, 4(1), 193-205.

27.

U.S. Bureau of Economic Analysis. (2015). Corporate non-residential fixed assets at current cost, excluding intellectual property (US Fixed Assets and Consumer Durable Goods, table 4.1). Washington, DC: Author.