Administering Health Plan Burdens During Economic Downturns: Lessons From Local Governments and the Great Recession

Abstract

The Great Recession exacerbated health care spending as a share of total revenue for local governments through the combination of decreased revenues and increased health plan premiums. This event forced the hand of local governments to choose between making layoffs, cutting services, or reducing employee benefits. Understanding more about why and how they did so is imperative for improving how local governments structure benefits during times of economic stability or surplus and also how they react during future economic downturns. In this study, a national survey of city and county administrators (N = 822) was conducted to understand if and how local governments introduced health plan burdens to their employees in response to the Great Recession. While a majority of local governments did not introduce significant health plan burdens to their employees, when a burden was introduced it was distributed disproportionately to current and future employees while protecting retired employees.

Keywords

health benefits local government Great Recession retired employee benefits unions

Introduction

The burden of sponsoring health benefits for current and retired local government employees has grown significantly over the past two decades. Health benefit liabilities for local government retirees grew so quickly in the late 1990s and early 2000s—now estimated to be $862 billion¹—that the Government Accounting Standards Board began requiring state and local governments to report all nonpension postemployment benefits and transition to prefunded models by 2007.^2,3 The burden of sponsoring health plans for current employees is similarly dire. In 2009, health benefits were cited as the second largest negative impact on city budgets, lagging only behind wages.⁴ City governments often report having to choose between reducing employees health plan benefits or cutting services such as public safety and human-social services.^4,5

Nevertheless, despite the rising costs of sponsoring employee health plans, state and local governments continue to offer them at more than double the rate of their private, for-profit counterparts—99.5% versus 47%, respectively.⁶ Not only do government employers offer more health benefits, but they also foot a greater percentage of the total premium. Whereas the average U.S. employer contributed $6,276 per employee in 2014, government employers contributed nearly 50% more—$9,289 per employee.⁷ Researchers have proffered reasons for this: benefit increases being less politically controversial than wage increases⁸; benefits standing in for lower-than-average public sector wages⁹; benefits being leveraged to attract, retain, and motivate employees,¹⁰ and the political clout of strong public employee unions.¹¹

These two phenomena—rapidly increasing sponsorship costs and ingrained institutional expectations—would collide during the Great Recession.^12,13 This event exacerbated the burden of sponsoring health benefits through the combination of decreased revenues and increased health plan premiums.¹⁴ This forced the hand of local governments to choose between making layoffs, cutting services, or reducing employee benefits.¹⁵

Though dire from an administrative perspective, the Great Recession provided a unique window into the priorities of local governments’ benefits managers. This article seeks to explore three questions about changes in health plan benefits using data collected from local benefits managers after the Great Recession. First, were health benefits scaled back in response to the Great Recession? Second, if health benefits were scaled back in local jurisdictions, how did the presence of unions affect those reductions? Third, if existing health plans were scaled back, how was this burden distributed across current, new, and retired employees?

Employee Health Benefits and Unions

Strong public employee unions are a key driver in the public sector’s benefits-to-wages ratio. Research demonstrates that unions can produce an annual wage and benefit premium of up to 10% for local government employees.¹⁶ Even unions without the right to collectively bargain can drive up wages and benefits.¹⁷ In the trough of the recession in 2009, nonunion state and local government employees were paid $3.07 in health benefits per hour worked versus $5.91 per hour worked for unionized workers.¹⁸ Over the past three decades, public sector union membership has remained high and not decreased, despite a downward trend in private sector union membership. In 2013, the public sector worker union membership rate was more than 5 times higher than their private sector counterparts—35.3% and 6.7%, respectively.¹⁹ This context raises an important subquestion for local governments looking to cut benefits in response to recessionary pressures: Since they have more to give, will union workers be forced to give up more in health benefits than nonunion ones or will unions protect them from conceding more health benefits than their nonunionized counterparts?

Distributing the Burden

Health benefits affect three pools of beneficiaries: retired, current, and prospective employees. Retirees—who are living longer and relying more on costly prescriptions and health care interventions²⁰—are a major source of concern for benefits managers.²¹ Even prior to the Great Recession less than 5% of local governments reported being able to prepay their “other postemployment benefits” liability.²² The difficulty in funding health coverage for retirees is multifold. For one, state and local governments have a history of offering retiree health coverage at a higher rate than the private sector.²³ Also, local governments comprise large numbers of public safety employees (e.g., police and firefighters), who retire before reaching age 65 and become Medicare-eligible. Offering health care coverage to this subgroup of retirees is twice as costly to employers than Medicare-eligible retirees: $9,064 versus $4,583. Moreover, whereas pension obligations are difficult to alter given their statutory and collective bargaining protections, retirees’ health benefits do not usually share the same protections and can be altered during financial crises.^24,25 Nonetheless, modifying health plan benefits for this pool has significant risks. First, cutting health benefits for pre-Medicare-eligible retirees reduces the labor force exit rate,²⁶ which reduces the impact of voluntary early retirement mechanisms.²⁷ Second, reductions in retiree health benefits go against the grain of societal expectations and can expect to be met with well-organized opposition.²⁸ Even with the Affordable Care Act’s regulated health insurance exchanges and lower age-band-rating ratios, pre-Medicare-eligible retirees face high out-of-pocket costs on the individual markets. Finally, it could be demoralizing for current employees facing their retirement futures.

Health benefits for current employees are the second largest pressure on local governments’ budgets.²⁹ Thus, given their generous benefit packages, making small cuts in health benefits would still leave local government employees with significantly more favorable coverage than their private sector counterparts.³⁰ If this option proves difficult to implement, scaling back health benefits for prospective (future) employees has been used by local governments during prior financial crises.³¹

However, targeting current and prospective employees also runs some risks: Retention of current employees and recruitment of new employees are heavily dependent on attractive health plan benefits.^32,33 Quality public employees, who often forego higher wages for historically superior fringe benefits,³⁴ could more easily choose the private sector. This context raises two important subquestions for local governments prioritizing health plan benefit cuts in response to recessionary pressures: Are the burdens represented by cuts in health benefits distributed evenly across job categories and what are the key determinants that account for any differences?

Method

Data Sources

Original data for this research were collected via a national survey of city (N = 382) and county (N = 440) managers and similar local governments officials on how they coped with the changes in local government budgets, service demands and operations after the economic downturn. The survey was administered between December 2012 and February 2013. The 822 total respondents replied in answer to e-mail solicitations sent to a list of 3,008 managers provided by the National Association of Counties and the National League of Cities. The response rate was 27.3%.

Survey Administration

The Florida Survey Research Center administered the online survey. Respondents were e-mailed requesting survey participation with an individually generated code number that allowed them to access the questionnaire online. Codes kept responses confidential and allowed for respondents to pause and continue answering later. Also, the codes allowed for tracking of responses both for follow-up and to separate unfinished questionnaires from finished ones. No unfinished questionnaires were included in the analysis. Two follow-up e-mails were sent to respondents who had not begun or not finished the questionnaire; the first follow-up was sent after 1 week and the second after 2 weeks.

Survey Instrument

The survey had 32 questions. Twenty-three were closed questions requiring multiple selections regarding the manager’s opinions on their jurisdiction’s general economic conditions, direction of economic change, and budgetary coping strategies, as well as reports on actual changes in service strategies and employee benefits. Of the 23 closed questions, there were 9 open-ended questions embedded to follow up if “other” was selected in response. Not all questions had the same number of substantive respondents due to the inclusion of a “prefer not to answer” response.

Dependent Variable: Health Plan Burden

Respondents were asked the following: “In the past 3 years, has your municipal/county government made any of the following changes to health care plans for current/new/retired employees?” (The question was asked three times—one for each category of job tenure.) Respondents were provided with the following six options and were encouraged to mark all that apply: decreased the city/county contribution level for health care plans, increased the city/county contribution level for health care plans, decreased the employee contribution level for health care plans, increased the employee contribution level for health care plans, reduced the number of health benefits covered and increased the number of health benefits covered. Each of the three responses that resulted in a greater employee burden was attributed a value of −1 (decreased city/county contribution, increased employee contribution and reduced total health benefits offered). Each response that suggested a reduction of employee burden was attributed a value of +1 (increased city/county contribution, decreased employee contribution and increased total health benefits offered). Each respondent’s selected options were then all summed to create a total burden score for that respondent. The more negative the burden score, the higher the burden that was placed on the employee. For example, a health plan burden score of −3 signifies that an employee would have experienced a reduction in total health plan benefits, a decrease in employer contributions, and an increase in employee contributions. Conversely, a health plan burden score of +3 signifies that an employee experienced an increase in total health plan benefits, an increase in employer contributions, and a decrease in employee contributions. A health plan burden score of 0 signifies that either no changes were made to health plan benefits or that one burden-increasing change was negated by one burden-decreasing change.

Independent Variables

Unionization

Unionization is operationalized as the percentage of public sector employees who belong to labor organizations.³⁵ Union membership percentages were taken from the Union Membership and Coverage Database³⁶ for 2012 and joined to the data set by state for each respondent. This cross-level joining is a common method used in the field when estimating union density for an entire city or county.³⁷

Past revenue

Respondents were asked to characterize their city’s/county’s revenues over the past 3 years (2009-2012). Options included: 1 = “declined a great deal,” 2 = “declined somewhat,” 3 = “stayed about the same,” 4 = “increased somewhat” and 5 = “increased a great deal” (see Note in the Conclusion section).

Current economic conditions

Respondents were asked to rate their city’s/county’s current economic conditions (approximately December of 2012). Options included: 1 = “poor,” 2 = “fair,” 3 = “good,” and 4 = “excellent.”

Control Variables

Controls were included for additional variables related to health plan offerings and the recession based on existing literature^38-40 including population, county government versus city government (0 = county, 1 = city) and region (1 = northeast, 2 = south, 3 = Midwest, or 4 = west).

Analysis

Differences in means of health plan burden scores across the three categories of job tenure are assessed through paired t tests. To examine the relationship between health plan burden and our independent variables, ordered logistic regression is used due to the polytomous nature of our dependent variables.⁴¹ Stata 13 was used to run all analyses. Coefficients, standard error and p levels are reported.

Results

Respondents in the study represent a wide range of populations. Twenty-five percent of respondents represent city or county populations of less than 25,000 and 10% of respondents represent city or county populations with more than 250,000 residents. City governments account for 47% of the respondents, with county and parish governments making up the difference. The northeast is notably underrepresented with only 54 of the 822 respondents. The other three regions are represented equally. Sixty-two percent of respondents indicate that their local governments’ revenues have either declined somewhat or declined a great deal over the past 3 years. However, only 20% report their current economic condition as poor. See Table 1 for additional descriptive statistics for the sample.

Table 1.

Descriptive Characteristics of the Sample and Independent Variables.

Sample characteristics	n	%
Population
< 5,000	39	5.2
5,000-9,999	43	5.8
10,000-24,999	94	12.6
25,000-49,999	189	25.4
50,000-99,999	185	24.9
100,000-249,999	123	16.5
250,000-499,999	39	5.2
≥500,000	32	4.3
City (vs. county)	382	46.5
Past 3 years of revenue
Declined a great deal	215	26.5
Declined somewhat	290	35.8
Stayed about the same	158	19.5
Increased somewhat	139	17.1
Increased a great deal	9	1.1
Current economic conditions
Poor	162	19.9
Fair	414	50.9
Good	217	26.7
Excellent	21	2.6
% Public union membership (state)
0-9	48	5.8
10-19	245	29.8
20-29	108	13.1
30-39	73	8.9
40-49	69	8.4
50-59	254	30.9
60-69	16	1.9
70-79	9	1.1
Region
Northeast	54	6.6
South	272	33.1
Midwest	240	29.2
West	256	31.1

Concerning the dependent variables, new and current employees experience relatively similar health plan burden distributions (see Tables 2 and 3). Twenty-one and 24% of new and current employees, respectively, receive two or more health plan burden points during the past 3 years. Meanwhile, only 8% of retirees receive this same level of burden. No respondents report improving health plan benefits by the maximum of 3 points and only two respondents (<1%) report improving health plan benefits by 2 points.

Table 2.

Observed Response Combinations and Health Plan Burden Score Derivations.

Detailed response	Resulting health plan burden score	New employees frequency	New employees %	Current employees frequency	Current employees %	Retired employees frequency	Retired employees %
No changes	0	379	46.1	238	29.0	593	72.1
Decreased employer contributions only	−1	19	2.3	25	3.0	24	2.9
Increased employer contributions only	+1	34	4.1	59	7.2	18	2.2
Decreased employee contributions only	+1	5	0.6	5	0.6	4	0.5
Increased employee contributions Only	−1	101	12.3	140	17.0	60	7.3
Decreased benefits covered only	−1	22	2.7	23	2.8	23	2.8
Decreased employer and employee contributions	0	5	0.6	6	0.7	3	0.4
Decreased employer contributions and increased employee contributions	−2	109	13.3	132	16.1	41	5.0
Decreased employer contributions and benefits covered	−2	5	0.6	5	0.6	5	0.6
Increased employer contributions and decreased employee contributions	+2	2	0.2	2	0.2	0	0.0
Increased employer and employee contributions	0	67	8.2	97	11.8	24	2.9
Increased employer contributions and decreased benefits covered	0	6	0.7	9	1.1	0	0.0
Increased employee contributions and decreased benefits covered	−2	18	2.2	24	2.9	13	1.6
Decreased employer & employee contributions and benefits covered	−1	0	0.0	2	0.2	0	0.0
Decreased employer contributions and benefits covered & increased employee contributions	−3	37	4.5	34	4.1	9	1.1
Increased employer contributions and decreased employee contributions & benefits covered	−1	0	0.0	0	0.0	0	0.0
Increased employer & employee contributions and decreased benefits covered	−1	13	1.6	21	2.6	5	0.6
Total		822	100	822	100	822	100

Table 3.

Descriptive Characteristics and t Test Comparison Across Dependent Variables.

Health plan burden score	New employees frequency	New employees M (SD)	Current employees frequency	Current employees M (SD)	Retired employees frequency	Retired employees M (SD)
−3	37		34		9
−2	132		161		59
−1	155		211		112
0	457	−0.597*** (0.972)	350	−0.691*** (1.014)	620	−0.286*** (0.684)
+1	39		64		22
+2	2		2		0
+3	0		0		0

Note. −3 = heaviest burden, 0 = no net change in health plan burden, +3 = improved contributions and benefits.

***

p < .01.

“No changes” in health plans receive the highest frequency across all three categories of employees: 46% for new employees, 29% for current employees and 72% for retired employees. Of interest for the subsequent analyses is that only a relatively small percentage of observations receiving a health plan burden score of 0 are derived via a combination of positive and negative health plan burden events. For example, of the 620 observations for retired employees that score a health plan burden of 0, a total of 593 (96%) are from respondents selecting “no changes.” That percentage falls to 83% for new employees and 68% for current employees. See Tables 2 and 3 for additional descriptive statistics for the dependent variables.

Results of paired t test comparisons of means across the three dependent variable pairings reveal significant differences for each (see Table 3). This indicates that current employees received the greatest health plan burden in response to the recession (M = −0.691, SD = 1.014) compared to new employees (M = −0.597, SD = 0.972) and retired employees (M = −0.286, SD = 0.684).

The univariate model results can be found in Table 4. Quite a few variable pairings are found to be significant for all three models. Thus, to check for multicollinearity, a variance inflation factor test was run and produced a mean variance inflation factor value of 1.18, well below the standard cutoff of 5 and signifying that multicollinearity does not present a serious threat to the model.⁴² Union membership and population are negatively associated with health plan burden across all three categories of job tenure. Declining past revenues and a negative economic outlook are also associated with a higher health plan burden, except in the case of retired employees. Also of note, city governments were more likely to report stronger past revenues, improved current economic conditions, and higher union membership percentages compared to county governments.

Table 4.

Correlations Matrix of All Noncategorical Variables Under Investigation.

Variables	New employee burden	Current employee burden	Retired employee burden	Population	City/county	Past revenue	Current economic conditions	% Unionized
New employee burden	1.0000
Current employee burden	.7599*	1.0000
Retired employee burden	.3145*	.3385*	1.0000
Population	−.1065*	−.1192*	−.1603*	1.0000
City/county	−.0473	−.0314	−.0885*	.2154*	1.0000
Past revenue	.1693*	.1486*	.0468	−.0325	.1504*	1.0000
Current economic conditions	.1001*	.0900*	.0331	.1208*	.2166*	.4339*	1.0000
% Unionized	−.2146*	−.1829*	−.0800*	.1163*	.1959*	−.1380*	−.1082*	1.0000

p < .05.

The ordinal logistic models produce similar results for new and current employees but tell a slightly different story for retired employees (see Table 5). For the former, each additional percentage point of unionization is associated with a .01 unit increase in the ordered log-odds of having a higher health plan burden (i.e., a more negative burden score), holding all other variables constant. New and current employee health plan burden scores are also both positively related to past revenue—a 1-unit increase in the reported past revenue scale is associated with a .27 and .23 increase, respectively, in the ordered log-odds of having a lower health plan burden score (i.e., a more positive burden score), holding all other variables constant. Finally, for new and current employees, health plan burden scores are negatively related to population—a 1-unit increase in category of residential population is associated with a .13 (for new employees) and a .14 (for current employees) increase in the ordered log-odds of having a higher health plan burden (i.e., a more negative burden score), holding all other variables constant. No other variables were significant at a p value of .05.

Table 5.

Ordinal Logistic Regression Results.

Variables	New employees	Current employees	Retired employees
% Unionized	−0.0119** (0.005)	−0.0105** (0.005)	−0.0090 (0.007)
Revenue past 3 years	0.2653*** (0.077)	0.2320*** (0.073)	−0.0142 (0.091)
Current economic conditions	0.1520 (0.112)	0.1491 (0.106)	0.1452 (0.134)
Population	−0.1258*** (0.047)	−0.1411*** (0.045)	−0.2273*** (0.058)
County (vs. city)	−0.1802 (0.161)	−0.1404 (0.155)	−0.2186 (0.192)
Region (northeast)	Reference	Reference	Reference
South	0.3979 (0.376)	0.2310 (0.352)	−0.5000 (0.453)
Midwest	−0.0600 (0.322)	−0.1646 (0.300)	−0.2972 (0.386)
West	0.2666 (0.328)	0.1808 (0.305)	−0.1357 (0.395)
Constant	6.1449 (0.881)	5.9199 (0.862)	2.2185 (0.660)
Observations	737	737	737
Nagelkerke R²	.083	.075	.043

p < .1. **p < .05. ***p < .01.

For retired employees only one variable—population—was significant at an alpha of .05. Its coefficient shared the same direction as it did for new and current employees—that is, the higher the population the higher the health plan burden (i.e., a more negative burden score). Our unionization and past revenue measures were not significant as they were for new and current employees.

Discussion

This article began with three objectives. First, to determine if local governments altered employee health benefits in response to the Great Recession. Our findings suggest that although many local governments did introduce some degree of health plan burden to their employees, most local governments avoided doing so. Sixty percent of local governments either did not introduce any health plan burden to new employees (a burden score of zero) or improved health benefits for them (a burden score greater than zero). That percentage fell to 50% for current employees but rose to 78% for retired employees. This finding coincides with existing cutback management research that suggests local governments often draw on reserve and savings funds during recessions⁴³ and choose other personnel-related cost containment measures, such as workforce reductions and layoffs, before directing their attention to health plan benefits.^44,45 While the merits of this strategy deserve additional attention, this study reconfirms that health plan benefits are not considered to be the lowest hanging fruit for cost-conscious government administrators.

Another possible explanation for the absence of health plan burden could be related to collective bargaining agreement durations. As most states in the United States allow for collective wage and benefit negotiations, these negotiations typically occur every 3 years, but in some cases longer.⁴⁶ Given that our dependent variable asked about changes in health plans “during the past 3 years,” it is possible that absence of a health plan burden might have been reported due to our 3-year study period overlapping with the 3-year duration of a collective bargaining agreement. However, if this circumstance was widespread in the data, one would not have expected 70% of respondents to report some level of change for current employee benefits, nor would one have expected the unionization variable to yield significant results, which it did in two of the three models.

After determining if local governments introduce health plan burdens in response to the recession, our second objective was to determine how local governments distribute said burden across retired, current, and prospective employees. As discussed, there are human resource concerns with penalizing any of our three categories of job tenure including recruitment, motivation, retention, and socionormative pressures. Our t test results suggest that current employees receive the greatest health plan burden compared to both new and retired employees. Seventy-two percent of respondents indicated that retired employees’ health benefits were neither harmed nor benefited in response to the recession. This finding suggests that retired government workers are likely benefitting from the socionormative pressures described in structured dependency theory⁴⁷ and are not considered open-season, save for the absolute worst of economic conditions. This is particularly encouraging when considering local governments’ large proportion of public safety workers who often retire before they reach age 65 and become Medicare-eligible. For these individuals, who are often on fixed incomes and, because of their age, consume more health care, any reductions in benefits or increases in employee contributions can result in severe financial impacts.

While the difference in the means of burden scores between current employees (−0.691) and new employees (−0.597) was statistically significant, the two means may not be meaningfully different given the polytomous nature of our dependent variable. In practice, the only differences in health plan benefits that employers make between current and new employees are the conditions of an employee’s future retirement health benefits (e.g., all employees hired after 2017 will contribute a larger percentage of the total premium than employees hired before 2016).⁴⁸

The third objective of this article was to look at factors that affect the shift of health plan burden to the three employee categories, notably, unionization. Unionization was significant in two of our three models, for new and current employees. These results suggest that just as unions achieve higher levels of benefits for employees during economic prosperity, so too do they give up more benefits during economic recessions. However, given the significant yet small coefficient, union workers are still likely to come out well ahead of their nonunion counterparts in terms of real health benefits per hour worked. This suggests that public sector unions are still weighty health plan protectionists in times of economic duress but do ultimately cede some additional ground compared to nonunionized counterparts. Though this finding is intuitive, it runs counter to another study that found unionized government employees received more benefits during times of budget shortfall.⁴⁹ However, that study relied solely on bivariate relationships.

Population was the only factor that was significant for all three dependent variables. The results indicate that larger cities and counties instituted more health plan burdens on new, current, and retired employees alike. This finding aligns with economic literature that suggests larger local governments fail to achieve economies of scale above a certain population threshold and end up with higher per capita expenditures.⁵⁰ Operating with these higher costs during severe economic downturns may call for above average cutbacks. This is a plausible explanation for the robust population-related results reported here.

Limitations

This study is not without its limitations. First, the dependent variables are simple integer constructs that do not use weighting or allow for much granularity. For example, an employee who experienced a significant increase in employee contributions but also a slight increase in employer contributions would have received a health plan burden score of zero. However, given that public sector wages are typically lower than private sector wages, this hypothetical employee did experience a health plan “burden” in that the net wages are reduced. To overcome this, a survey would have required respondents (or researchers) to laboriously report how the myriad health plans offered had changed in the study years, which would have resulted in a dismal response rate. Still another limitation of the dependent variable is that it did not capture instances of employees having a higher or lower quantity of plans available to them. While a reduction in choice from, say, three health plans to two may not be particularly burdensome for employees, it can be disruptive for those enrolled in the particular plan that is discontinued by an employer. Fortunately, our survey did catch any instances of employers going from offering a single plan to no plan at all.

Second, significant proportions of the model variance in health plan burden remain unknown. Our highest R² statistic is .086 for the current employee model, indicating that the model only predicts 9% of the variability in health plan burden. This could be due to factor omission, measurement issues, or that local governments respond randomly and locally to severe budget constraints—a theory demonstrated in other research.^51-53

Our unionization variable is a state-level variable that was joined to and used as proxy for local-level data. This can result in high-union cities and counties in low-union states to be underinflated and low-union counties and cities in high-union states to be overinflated. However, this cross-level joining is a common method used in the field when estimating union density for an entire city or county.⁵⁴

The total number of respondents from the Northeast was lower relative to other regions. However, this is likely due to inclusion of county data and how U.S. Census Regions are defined (the South includes 1,380 total counties vs. 217 in the Northeast). Further analysis revealed that county response rate was higher in the Northeast (13%) than the South (11%).

Finally, our analysis does not account for fiscal capacity—that is, a government’s ability to deal with the Great Recession. Fiscal variables like cash reserves and fund balances are key determinants of how local governments respond to a recession. While this article includes a constructed measure of the Great Recession’s impact on a local government’s revenues, it does not include the government’s capacity to moderate the effect of revenue change on employee benefits. Future studies could use administrative data to estimate how fiscal capacity protects employee benefits.

Conclusion

Despite the growing financial burden of sponsoring generous health plans for their employees, this study suggests that U.S. local governments avoid cutting employee health benefits significantly during a financial crisis. The Great Recession—the worst global recession since the Second World War—tested local governments’ fortitude in upholding this practice. Yet, for the most part, they held the line and weathered the storm without cutting employee benefits. A majority of this study’s local governments reported that they did not cut benefits.

Nevertheless, some local governments did show a greater propensity to shift health plan burdens to some employee groups. These were the larger, more populous local governments that experienced more dramatic 3-year revenue losses and, importantly, were more highly unionized. These strained local governments shifted the greatest health plan burden to current and new employees and, for the most part, left retirees’ health benefits untouched.

This fact cuts two ways. First, it may signify that larger jurisdictions with larger budgets and workforces have fewer places to look for savings. Instead, they must work at the budgetary margins and shift benefit costs where possible rather than cutting them. Too, this is dictated by the political makeup of larger jurisdictions that often mitigates against drastic cuts for labor. Second, it may indicate that the power of employee unions in larger jurisdictions is waning to some degree: Greater employee burdens for health care affect both the paychecks of current union members and make union representation less attractive to and important for recruiting new employees; union ability to protect benefits is constrained. Rather than underscoring union power, the failure of these local governments to shift burdens to retired employees may be both a reputational issue—keeping a promise in a completed agreement—and a strategic choice: to make financial changes that will affect budgets over the long run by changing the financial calculus for most current and all future employees. If the latter is the case, it may imply waning influence for some local government unions.

Note: In an effort to reduce single source bias, we conducted the analysis with two different measures of this variable: once using this self-reported measure of past revenue and once using data from the Government Finance Database.⁵⁵ For the latter, we used an own-source total revenue measure, as we wanted to exclude state and federal intergovernmental revenues, which may have included stimulus monies unique to the recession. We created a proxy measurement of the impact of the recession by calculating the percentage change of own-source total revenue from 2008 to 2011 for each county government observation, to align with the survey question about the previous 3 years of revenue. However, as 2008 and 2011 were not years in which a full census was conducted, only 51% of our observations included data for both 2008 and 2011—a known issue with the U.S. Census Bureau data.⁵⁶ To maximize the match rate, one would need to use 2007 and 2012 data—the years in which a full census was conducted. However, these two observation years not only do not match up with our survey question but also preclude us from measuring a more granular impact of the Great Recession, which officially began in December of 2007 and ended in June of 2009. Thus, 2007 and 2012 data are likely only capturing the bookends of the recession and do not allow for meaningful imputation. Regardless, we conducted the analysis using only the county observations that included 2008 and 2011 data on own-source total revenue (n = 226) and compared them with the self-reported measure of past revenue. Using the U.S. Census–derived measure, the directionality was the same for all independent and control variables, though the measures for past revenue and unionization were no longer significant (p < .05). Only population remained significant (p < .05). This finding, combined with a much improved sample size, convinced us to use the self-reported measure of past revenue.

Footnotes

Declaration of Conflicting Interests

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

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Nicholas Edwardson is an assistant professor at the School of Public Administration and a Senior Fellow at the Robert Wood Johnson Foundation Center for Health Policy at the University of New Mexico.

Bruce J. Perlman is Regents’ professor and Director of the University of New Mexico’s School of Public Administration. Among other positions, he has served as Chief Administrative Officer of the City of Albuquerque and Deputy Chief of Staff in the office of the Governor of New Mexico.

Michael J. Scicchitano is the director of the Florida Survey Research Center at the University of Florida. He is also Editor in Chief of State and Local Government Review.