Printer Friendly

What factors affect the underfunding of local pensions? Evidence from Indiana.

1. INTRODUCTION

Underfunded local government pension plans place a burden on future taxpayers for public services performed in the past. Not surprisingly, unfunded pensions at the state and local level have become a critical public policy issue, so are worthy of closer scrutiny than they have thus far been afforded. Pension obligations were a significant factor in the re99cent bankruptcies of Vallejo, California, Central Falls, Rhode Island, and Detroit, Michigan (Dye and Gordon 2012, 3). Pension underfunding leads to lower credit ratings and thus, to higher borrowing costs. Pension underfunding masks the true cost of current services and encourages increased spending. It also violates interperiod equity by shifting current costs to future taxpayers, and it may decrease property values. Recent changes in governmental accounting standards may render pension funding shortfalls more visible, but will not necessarily lead to solutions. The purpose of this research is to identify factors contributing to pension underfunding among Indiana's local governments. The results will help local officials avoid pension funding crises and achieve healthy funding levels.

This study extends the literature in the following ways. First, using data from Indiana we examine the determinants of underfunding of local pensions. Due to data constraints, most of the previous research has examined state pension programs. While important, this does not clearly outline the public policy dimension of underfunded local pension plans. Also, much of the analysis of pension plans has focused on the size, the actuarial certainty, the impact of underfunding on borrowing costs and the relationship between pension funding levels and the composition of debt. In this work, we examine factors that potentially contribute to underfunding of defined benefit plans. Recent changes in governmental accounting standards will increase pressure on local governments to more fully fund their pension plans, and our results will help them respond effectively.

Our study's focus is the funded status of local public pension plans. We use data from Indiana to examine the determinants of underfunding. We find that economic and demographic factors such as per capita income and migration, along with fiscal factors such as the average tax levy and state aid, and broad interest group influences such as per capita retirement income are the major determinants of pension underfunding among local governments in Indiana.

In the next sections we review the literature on defined benefit plans, the funding levels of public pension plans, effects of underfunded pensions and factors that influence underfunding. We then discuss the hypotheses that we tested related to local pension underfunding. Next, we discuss the data and empirical method used to test these hypotheses followed by a discussion of results. The final section offers conclusions and discusses implications and limitations.

2. DEFINED BENEFIT PENSION PLANS IN THE PUBLIC SECTOR

Pensions are compensation for work performed, and thus, shift a portion of labor costs from the present into the future. A defined benefit plan promises a certain level of monetary benefits to workers upon retirement, usually in proportion to their years of service and final wage or salary. The promise is independent of the resources actually set aside and held in trust to fund the future benefits, though employers usually make regular contributions to pension funds (Ippolito 1985, 612). A defined benefit plan is underfunded if the value of plan assets is less than the plan's actuarial accrued liability (Vermeer et al 2012, 47). Some public pension plans are so underfunded that it may precipitate a crisis for the government employer and for the retirees, who may not receive their promised benefits.

While defined benefit plans are quickly disappearing from the private sector, about 80% of state and local government employees are covered by such plans (U.S. Government Accountability Office (GAO) 2012, 5; Dye and Gordon 2012, 4; Rich and Zhang 2014, 1). The United States has more than 3,400 state and local pension systems, covering about 27 million state and local workers and retirees (GAO 2012, 1). Local governments often participate in state-administered plans, yet contributions into state-administered plans often remain a local decision. Most public school teachers are covered by state-administered plans, but police and fire workers are more likely to be covered by local pension plans (Inman 1982, 51).

Determining the funded status of a plan is not a straight-forward proposition. Complex, actuarial assumptions are used to calculate required contributions to pension plans as well as the funded status (ratio of plan assets to projected pension obligations) at any point in time. The net pension liability is the difference between accumulated assets and the actuarially determined liability (Mortimer and Henderson 2014, 431). Critics charge that in the past, the Governmental Accounting Standards Board (GASB) contributed to confusion over funding status through poorly-conceived accounting standards. For instance, GASB previously allowed a choice of six different actuarial methods for projecting future obligations and allowed use of overly-optimistic rates for discounting future liabilities (GASB 2012, 45; Dye and Gordon 2012, 5). The new GASB pension accounting standard, Statement No. 68: Accounting and Financial Reporting for Pensions (1), became effective for fiscal years beginning after June 15, 2014; it requires use of a single actuarial method and a more conservative, composite discount rate. Moreover, it requires governments to report the net pension liability of defined benefit plans on the face of the balance sheet, thus ending the off-balance sheet treatment of these liabilities (Rich and Zhang 2014, 5; GAO 2012, 47: GASB 2012, 9).

The funded status of a defined benefit plan is affected by several factors. These include the investment return on plan assets; the level of annual contributions; deviations from actuarial assumptions (e.g. workers living longer than expected); an increase or decrease in promised benefits; and the choice of the discount rate to discount future liabilities (Munnell et al, 2015). While the new GASB standard will improve pension reporting, assessing the funded status of a pension plan remains an exercise in professional judgment; no single benchmark will indicate a healthy or unhealthy plan. The "80%" criterion, referring to plans where assets held in trust equal at least 80% of actuarially projected obligations, has been widely cited but recently debunked as a single, reliable measure of a healthy plan. The American Academy of Actuaries (AAA) and the Government Finance Officers Association (GFOA) argue that all pension plans should be managed to achieve a funding status of 100% or greater over a reasonable period of time (AAA 2012, 1-2; GFOA 2015). (2) Additional factors, such as the contribution record or the size of actuarially required contributions in relation to total revenues or total payroll, should be considered along with funded ratios when assessing the health of a plan (AAA 2012, 3).

Recent practice (prior to implementation of GASB Statement 68) reveals considerable variation in choice of actuarial methods and discount rates, and pension obligations are easily manipulated by the choice of discount rate (3) (Marks et al 1988, 160). Governments vary significantly in their financial skills and level of disclosure (Farmer 2015, 57), and many governments do not follow GASB standards (Khumawala et al, 2014).

3. FUNDED STATUS OF PUBLIC PENSION PLANS

While there is considerable variation among plans, many public pension plans are seriously underfunded (Collins & Rettenmaier 2010, 9; Dye and Gordon 2012, 8-9; Farmer 2015, 57). Some plans are so underfunded that to bring them to a fully funded status would require more than 100% of current payroll (Dye and Gordon 2012, 8). The Pew Center on the States estimates that state pension plans have a combined shortfall of $757 billion and local plans have a combined shortfall of $9 billion (Lambert 2013). Dye and Gordon report that "A significant fraction of local governments are in trouble" (2012, 9). Rauh concludes that based on current estimates, pension fund assets may be exhausted within 15 years (2011). Kelley analyzed self-reported data for the 126 state and local plans found in the Public Pensions Database (4) and finds the average state has about $1,280 in unfunded liabilities per capita (2014, 29-30). The GAO (2012, 7) is less alarming but still cautionary; it finds that most state and local plans can cover their commitments "for a decade or more" but that higher contributions are required if these plans are to remain sustainable.

The higher the discount rate, the lower the present value of future obligations. In the past, plan administrators typically chose discount rates around 7% to 8% (Novy-Marx and Rauh 2011, 54). Recently, scholars have recalculated projected pension obligations using discount rates similar to those required under GASB 68 and the result is a deterioration in funded status. Collins and Rettenmaier (2010, 451) recalculated the liabilities of 153 state and local pension plans and found their unfunded pension liabilities to be $2.5 trillion rather than $493 billion as reported under previous assumptions; average funded ratios decreased from 73.4% to 56.3%. The journal Governing analyzed 80 state and local pension plans and found that under the new GASB guidelines, the average increase in projected liabilities was only nine per cent but for some plans the increase was as high 55% (Farmer and Maciag, 2015). Mortimer and Henderson studied 48 major, state-administered defined benefit plans; for their sample, the average funded ratio decreased from 73.4% to 56.3% under the new guidelines (2014, 439).

Some analysts argue that discount rates should be based on U.S. Treasury yields since public pension systems are virtually "risk-free" (Rich and Zhang 2014, 15; Mortimer and Henderson 2014, 450; Rauh 2011). Novy-Marx and Rauh (2011) analyzed 77 local pension plans, including all municipal entities with more than $1 billion in pension assets. When using a discount rate based on high-quality municipal bonds, aligning with the new GASB standard, the aggregate pension liability for their sample rises from $581 billion, as currently reported, to $662 billion (Novy-Marx and Rauh 2011, 63); when using a discount rate based on U.S. Treasury yields, the aggregate pension liability rises to $1.047 trillion.

Periods of economic contraction increase fiscal stress because tax revenues decline simultaneously with an increased demand for government services. The recent "great recession" increased fiscal pressure on state and local governments, making it harder for them to meet pension funding requirements. As revenues fell, the value of pension fiduciary assets also fell, which increased the actuarially required amount of contributions (GAO 2012, 17). During 2008 and 2009, state and local pension plan investments lost more than $672 billion (GAO 2012, 8). Too often, when contributions are deferred to close a current budget gap, they are not fully restored in future periods, and some local governments borrow from pension funds to cover immediate needs (Chaney et al 2002, 288; Lyons & Lav 2007, 7).

4. EFFECTS OF UNDERFUNDED PENSION PLANS.

Large unfunded pension obligations lead to lower credit ratings (Dye and Gordon 2012, 3; Marks et al 1988, 159). Between 2010 and 2013, Moody's (2013, 5) downgraded the credit ratings of Illinois, Connecticut, Kentucky, New Jersey, Hawaii and Pennsylvania, due largely to growing pension liabilities. Moody's also reviews local governments for possible downgrades because of pension gaps (Lambert 2013). In 2012, Moody's downgraded Chicago's outstanding general obligation debt from stable to negative, due in part to underfunded pension liabilities (Rich and Zhang 2014, 1). Lower credit ratings lead to higher borrowing costs. Rauh (2011) finds that tax-adjusted municipal bond yields rise by ten to 20 basis points for each decline in pension plan funded status equal to 1% of annual, gross state product.

Ultimately, unfunded pension liabilities can lead to default and bankruptcy, as happened recently in the cities of Vallejo, California, and Central Falls, Rhode Island (Dye and Gordon 2012, 3). In earlier years, pension liabilities were major factors in the fiscal crises of New York City and Cleveland (Inman 1982, 50). Public employee pensions have strong legal protections that are not easily removed (Munnell et al 2008, 1; Kelley 2014, 23), yet more jurisdictions may turn to bankruptcy as a possible strategy for reducing unsustainable pension obligations (GAO 2012, 18; Geiger et al 2015; Kelley 2014, 24).

A less visible outcome is the incentive to overspend on current government services. Lack of visibility into true, current costs may lead to an "inefficiently high" level of services (Marks et al 1988, 159). Inman finds that as the funded status of police and fire fighter pensions falls, communities hire more police and fire fighters (Inman 1982, 68). Research also shows that states with greater pension deficits are more likely to hire additional employees, grant more generous retirement packages, and have higher levels of general expenditures (Rich and Zhang 2014, 5).

Another potential outcome is decrease in property values. Because pension payments are "guaranteed" through property taxes, future taxpayers who purchase homes in the community may require compensation for assuming this burden in the form of lower property costs (see Rich and Zhang 2014,, 8; Epple and Schipper 1981, 142; Inman 1982, 55). Further, large unfunded liabilities may signal potential buyers to choose another community, again lowering property values (Rich and Zhang 2014, 8).

Unfunded pension liabilities violate interperiod equity (IPE), defined by GASB (1987, 18-19) as "whether current-year revenues are sufficient to pay for the services provided that year and whether future taxpayers will be required to assume burdens for services previously provided." Marks and Raman (1996, 64) argue that governments need time to adjust to changing economic circumstances and, therefore, IPE must be achieved "over a multi-year time frame rather than each and every year." Still, a persistent pattern of underfunding pension obligations represents a de facto subsidy from future to current taxpayers (Inman 1986, 21; Dye and Gordon 2012, 9). Unfunded liabilities are more than a short-term problem; Rich and Zhang estimate that if current pension deficits were to be eliminated over a period of 30 years, the increased annual tax burden would be $1,385 per household (2014, 1).

Faced with large unfunded pension liabilities, governments may invest in high-risk portfolios, hoping to achieve higher returns on pension fund assets (Dye and Gordon 2012, 6; GAO 2012, 26). Another high-risk response is to issue pension obligation bonds; the proceeds enrich pension assets, yet the bonds create new, inflexible liabilities making it even more difficult for the government to adjust to economic changes in the future (GAO 2012, 19; Moodys 2012, 5). Such "artificial" solutions can make the pension system seem healthier than it is, reducing incentives for reform (Moodys 2012, 6).

5. FACTORS THAT IMPACT PENSION FUNDING LEVELS.

Multiple factors contribute to the underfunding of defined benefit pension plans. We examine three broad categories to analyze the determinants of underfunding.

Prior research finds a relationship between demographic factors and pension funding levels. Per Inman (1986, 29), the most underfunded local plans are in poorer rural states or in older, more industrialized cities such as New York City, Detroit, and Chicago. Funding ratios trend lower in jurisdictions that have been losing people and jobs over time; examples are Detroit, Michigan, which has twice as many public pensioners as active public employees, and Prichard, Alabama, which has lost more than 45% of its population since 1970 (Dye and Gordon 2012, 3).

Another explanation is the "mover" model (Inman 1982; Johnson 1997; Courant et al 1979; Novy-Marx and Rauh 2011). If taxpayers expect to remain in the community, there is less incentive to underfund pensions because they will not avoid paying for promised pensions. Taxpayers who expect to leave, however, have an incentive to delay payment for current services by passing on pension obligations to the residents left behind or to new resident who will replace them. Johnson (1997, 115) finds a strong, positive relationship between the level of underfunding among state-administered pension plans and the fraction of the population leaving the state over a five-year period. The "mover" effect is mitigated if unfunded pension obligations are capitalized into lower land values (Johnson 1997, 116). Also, the mover phenomenon, or threat of out-migration, might work to constrain excessive pension promises (Courant et al, 1979, 817).

Pension funding levels are related to the local fiscal environment. Measures of fiscal stress include current year general fund surplus (or deficits), cumulative general fund balance, changes in unemployment rates, long-term debt, and bond ratings (Chaney et al 2002. 295-297). Rich and Zhang (2014, 17) find that per capita debt varies directly with underfunding of municipal pension plans, and Johnson (1997, 130) finds a positive association between debt burdens and underfunding levels for both state and local plans. Marks et al (1988, 176) caution that the association between fiscal stress and pension funding is bi-directional; attempts to more fully fund pensions may increase fiscal stress on the government.

Fiscal stress is exasperated through balanced budget requirements and property tax caps (Rich and Zhang 2014, 5). Deferral or reduction of pension contributions is a common technique for meeting balanced budget requirements; states that are fiscally stressed and have balanced budget requirements fund their pensions at significantly lower levels than other states (Chaney et al 2002, 288). "Tax caps" restrict property tax increases from year to year; caps are usually based on a low percentage rate, the inflation rate, or some combination of the two (Lyons and Lav 2007, 1). A rigid tax cap inhibits raising the amount of pension contributions to meet actuarial requirements; either the pension contributions are deferred or other services must be cut (Lyons and Lav 2007, 8). Inman finds that tax limits reduce the funding levels of police pension plans, though the effect is small (1982, 64).

Researchers have also found that state intergovernmental revenues depress, rather than raise, own-source contributions to municipal pension plans (Inman 1982, 64). Local officials may interpret state aid as a signal that states will come to the rescue of underfunded pension plans (Inman 1982, 64; Novy-Marx and Rauh 2011, 56). The results of these studies suggest that unfunded liability should be higher in places receiving more state intergovernmental aid.

Scholars have applied a public choice perspective, particularly the median voter hypothesis and the role of special interests, to explain pension funding levels. Funding decisions are driven not merely by rational actuarial and managerial factors; political factors and interest groups play a role (Kelley 2014, 21). Public employees can affect pension promises and funding levels both as voters and as members of an interest group, even without formal collective bargaining. If the public workforce is large enough or well organized, it can shift compensation levels and funding decisions away from the preferences of the median voter (Courant et al 1979, 815; Kelley 2014, 22). In support of the special interest hypothesis, using data on the 126 largest state and local public pension systems in the United States, Kelley (2014, 35) finds that the proportion of retired public pensioners to total state population is a significant determinant of unfunded liabilities. Due to their shorter time horizons, current retirees face less risk from underfunded pensions. Conversely, a larger proportion of younger participants reduces unfunded liabilities because they will incur the future consequences of underfunding (Kelley 2014, 35).

As a special interest group, public employee unions could have variable impact on pension funding. Unions could use their influence to encourage adequate funding of pension systems to ensure their members receive the deferred benefits. Alternatively, they could use their influence to exact more generous benefit terms than the government can realistically afford, resulting in lower funding ratios. Prior research has identified both effects. Using annual pension reports filed with the U.S. Department of Labor between 1960 and 1981, Ippolito (1985, 634 and 638) finds that pension payments for unionized employers are almost 50% higher than payments for nonunion workers, yet the funded ratios for union plans "are more than 30% lower than nonunion plans." Ippolito explains his findings in terms of union "hold-ups" and bonding theory. Using their collective power, unionized workers can demand such high levels of compensation as to threaten the long-term survival of the firm. To counteract this threat, firms (and governments) deliberately underfund pensions so that workers have a stake in the firm's long-term survival, thus mitigating the hold-up effect. Workers are "bonded" to the firm through underfunded pensions (Ippolito 1985, 611-617). Marks et al (1988, 176-178) agree that unionization leads to lower funding levels and that "pension underfunding may have a 'bonding' role in the public sector." Recently, however, other scholars find that greater union presence is associated with better funded pension plans (Rich and Zhang 2014, 19; Chaney et al 2002, 307; Kelley 2014, 33). Munnell (2012) reports "It is impossible to find a link between a plan's generosity or funding level and the strength of unions."

Vermeer et al (2010, 532) find that sole-employer pension plans with greater participation of police and fire workers adopt less optimistic actuarial methods and assumptions; the less optimistic assumptions do not guarantee higher funding ratios, but do make it harder to mask future obligations. In contrast, Rich and Zhang (2014, 31) find no statistically significant difference in funding levels for uniformed employees. They do report a negative correlation between underfunding and the covered payroll per capita, implying that plans with more participating employees are less underfunded (Rich and Zhang 2014, 17). We included vested employees as a share of local government employment to measure special interest effects, but it may also capture this later effect. As described below, the pension data used in the empirical analysis includes mainly police and fire pensions and so may represent a lower bound in terms of underfunding.

We include variables that are commonly used to test median voter hypothesis in the model described below. However, the results of some studies suggest that the median voter may not be a viable determinant of pension funding. The complexity and opacity of public pension systems, especially prior to implementation of GASB 68, create "fiscal illusion" and make it easy to hide the true cost of current service from voters (Marks et al 1988, 167; see also Kelley 2014, 26). Further, complexity and opacity create high information costs so it is rational for voters to remain ignorant (Kelley 2014, 25). A pension task force of the U.S. Congress House Committee on Education and Labor (1978) finds a "high degree of pension cost blindness" among plan participants, plan sponsors, and the general public (1978, 4). Under the circumstances described in these studies the median voter is not likely to influence pension underfunding.

Additional factors that impact pension funding levels include quality of financial disclosure. One study finds "remarkable" consistency between the quantity and quality of financial disclosure, such as conformance with generally accepted accounting principles (GAAP) promulgated in GASB standards, and "lower magnitudes of underfunding" (Marks et al 1988, 179).

6. HYPOTHESES

Focusing our analysis on the determinants of local government unfunded pension liabilities, we test three hypotheses considered in the literature. The first is that declining economic conditions and corresponding demographic changes decrease funds available for pension liabilities leading to greater unfunded liabilities. The second hypothesis is that fiscal conditions due to fiscal policy changes or differences in fiscal structures contribute to underfunded pension plans. The third hypothesis is that special interests exert pressure to maintain higher levels of public services, while deferring costs to future taxpayers. We use data on local pension plans in Indiana to test these hypotheses.

7. DATA AND EMPIRICAL METHOD

We aggregate FY 2013 unfunded local government pension liabilities to the county level in Indiana's 75 counties with defined benefit plans. (5) Data is derived from the Indiana Gateway for Local Government Units (Indiana Gateway) and includes 158 separate defined benefit pension plans provided by Indiana local governments (county, city and special districts). This data is prior to the full implementation of GASB Statements No. 67 and No. 68, and so likely underreports both the number and size of unfunded pension liabilities in Indiana, based upon those criteria.

The majority of the defined benefit plans included in the Indiana Gateway are public safety (police/sheriff) pension funds with some sanitary and hospital employee pension funds also included. (6) Figure 1 shows the distribution of unfunded liability and the total funding ratio among Indiana counties.

We then construct a simple empirical model with data to test each of these hypotheses. We test whether the size of the unfunded liability within a county's local government pensions is associated with the state of the local economy, differences in the fiscal environment, or the influence of interest groups.

Some special modeling considerations are warranted in this setting. The distribution of unfunded liabilities is not normal, or a random event; some counties exhibit no unfunded liabilities, while the largest share of unfunded liabilities and the individual funds with the largest funding deficits occur in just a few counties. See and figure 1 for more detail. To model this we use the extreme value distribution with White/Huber standard errors to correct heteroskedasticity (White, 1980). (7) The basic model takes the following form

[U.sub.i] = g([E.sub.i], [F.sub.i], [I.sub.i]).

In this expression, [U.sub.i] represents the measure of unfunded pension liability, g is the functional relationship of a generalized extreme value distribution, [E.sub.i], are economic characteristics of county i, [F.sub.i] are fiscal characteristics of county i, and [I.sub.i] are interest group characteristics of county i. Table 1 lists the variables used in the model and sources. We estimate two variations of the model.

We include two measures of unfunded pension liabilities: unfunded actuarial accrued liability per capita and unfunded actuarial accrued liability per current local government employee. The two variations provide measures of the unfunded liability per resident of the county and per local government employee in the county and also provide a robustness check for the model. While the results are similar between the two models Akaike information criterion (table 3) shows that the model with underfunding per capita better fits the data than does the per employee model. However, both models should be interpreted as providing robust estimates. See Long (1997) for the application of the AIC.

[FIGURE 1 OMITTED]

Changes to economic conditions that might lead to budgetary shortfalls and unfunded pensions offer the most intuitive explanations for pension underfunding. We use measures of economic wellbeing such as per capita income and unemployment rate, along with an urban measure (the MSA variable) and the presence of a state border. This later variable could arguably include fiscal conditions because the presence of a state border might alter tax flows and also influence migration flows. We also include the percent change in the county population from 2003 to 2013 to capture long-term population changes.

There is much variation in population trends within the state with some counties experiencing large increases in population and others large declines. See descriptive statistics in table 2. We expect there to be a negative relationship between the population growth rate and unfunded pension liability. Population decreases are expected to lead to revenue declines and higher unfunded pension liabilities. We also include measures of domestic outmigration, the number of people moving out of the county to other counties within Indiana or other states between 2008 and 2012, and domestic in migration, the number of people moving into the county from other Indiana counties or other states over the same period.

Previous analysis (Johnson 1997) found that states with more outmigration have higher levels of unfunded pension liability but did not control for in migration. Courant et al (1979) suggests that outmigration may constrain unfunded liabilities.

Next, we examine the relationship between the local fiscal environment and unfunded pension liability. The property tax, local income tax and intergovernmental transfers from state government are the primary sources of revenue for local governments in Indiana. The per capita net levy and the effective tax rate on county residents are measures of the size and relative share of taxed property value available to meet pension fund obligations. Likewise the local income tax rates measures potential revenue from this source. Also, the population adjusted reduction in property tax revenue due to property tax caps is included as a measure of fiscal stress.

Finally, our proxies for interest group variables include the per capita level of income in retirement earnings in a county which reflects the broader influence of retired workers in a community, the employment share of state and local government workers in a county, current local government retirees as a share of the local government workforce and current vested local government employees as a share of the local government workforce. These variables capture the influence of each of these groups. (8) The theory at work here is that unfunded pensions defer the cost of current public services to future taxpayers. Voters may prefer this because it may reduce the lifetime costs of government services by shifting costs to future households. Because unfunded pensions defer the cost of current public services to future taxpayers, some of these interest groups may seek to reduce their lifetime cost of government services by shifting costs to future households.

8. RESULTS

The model results (table 3) provide limited support for the three hypotheses tested. (9) The hypothesis that a troubled local economy contributes to unfunded pension liabilities is supported by the statistical significance of the income variable. Communities with lower per capita incomes experience greater unfunded pension liabilities. The unemployment rate and population change, which are more direct measures of economic distress, are not significant.

Migration does influence pension underfunding but not in the expected way. Counties with larger shares of recent population outmigration are associated with lower levels of underfunding per capita supporting Courant's point that outmigration may constrain unfunded liabilities rather than the mover model (Johnson 1997) which suggests that mobile segments of the population are unlikely to bear the burden of paying pension obligations because they can move, while less mobile segments of the population (often, low income households) will bear the burden. In migration, which has not been tested in previous studies, is associated with higher levels of underfunding which suggests that local pension obligations are not a particular concern for movers.

There is evidence that a county's location within the state affects pension underfunding. Counties that border other states are associated with lower levels of pension underfunding which suggests structural differences in the local economy of border counties relative to interior counties. Importantly, these variables could also indirectly influence fiscal variables. However, the urban measure did not have statistical meaning of consequence suggesting that urban places are not more prone to pension underfunding in Indiana.

Only two of the fiscal variables have significant impacts on unfunded pension liabilities. Higher average property tax payments (net property tax levy per capita) have a positive association with unfunded liabilities. Because places with higher average levies have higher levels of unfunded pension liability, this likely precludes tax increases as an option to eliminate unfunded liabilities. The model shows a negative relationship between state intergovernmental revenue and unfunded pension liability, which suggests that unfunded liabilities are lower in places with more state intergovernmental aid. This result is different from previous studies (Inman 1982 and Novy-Marx and Rauh 2011), and given fungibility within local budgets likely indicates that nonresidents bear a portion of the burden of local pensions. Neither property tax rates nor local income tax rates are significant determinants of pension underfunding.

Only two of the fiscal variables have significant impacts on unfunded pension liabilities. Higher average property tax payments (net property tax levy per capita) have a positive association with unfunded liabilities. Because places with higher average levies have higher levels of unfunded pension liability, this likely precludes tax increases as an option to eliminate unfunded liabilities. The model shows a negative relationship between state intergovernmental revenue and unfunded pension liability, which suggests that unfunded liabilities are lower in places with more state intergovernmental aid. This result is different from previous studies (Inman 1982 and Novy-Marx and Rauh 2011), and given fungibility within local budgets likely indicates that nonresidents bear a portion of the burden of local pensions. Neither property tax rates nor local income tax rates are significant determinants of pension underfunding.

Property tax credits stemming from property tax caps, which is a measure of the amount of revenue local governments forego due to property tax caps and a measure of fiscal stress, are not a significant determinant of unfunded liabilities likely due to the recent implementation of tax caps. This suggests that property tax caps are not a causative element in unfunded liabilities in Indiana (at this point). These results suggest that fiscal factors involved with the management of funds, not restrictions on local government revenues dominate the unfunded pension financing issues.

Next we examine the relationship between interest groups and unfunded pension liabilities. Only one of the interest group variables is statistically significant: retirement income per capita, which is the broadest measure of influence of retirees used in the model. We find that counties with higher levels of retirement income have lower unfunded liabilities. This result suggests that retirees serve to constrain pension underfunding, which provides indirect support for the interest group hypothesis. The negative signs on the local government share of total employment and share of vested local government employees suggest that these employees use their influence to promote a healthy pension system, while the positive sign on current local government retirees suggests that these employees are more likely to push pension obligations into the future because face less risk from underfunded pensions due to a shorter time horizon. However, these variables are not significant determinants of underfunding; hence these results do not support the hypothesis that interest groups exert pressure to defer pension costs to future taxpayers.

There are three clear interpretations from this model. First, economic distress (as measured by income) is a contributor to unfunded local government pensions in Indiana. Second, it is budgetary decisions within city and county councils more than available revenue that influences the fiscal contribution to pension plans. In particular, counties with higher average tax levies and lower state intergovernmental revenue have larger unfunded liabilities in Indiana. Finally, there is not a strong relationship between the typical measures of interest group influence, such as local government share of employment and share of local employees who are vested, and pension underfunding. We find that the broad influence of retirees (all retirees not just local government retirees), plays a role in constraining unfunded liabilities.

9. CONCLUSIONS, IMPLICATIONS AND LIMITATIONS

Adequate pension funding continues to be an issue of concern for local governments. Pension obligations can dramatically affect the fiscal health of local governments and has implications for the quality of local services and tax obligations of residents and businesses. New reporting standards seek to quantify and to more clearly communicate the magnitude of underfunding.

The statistical analysis presented in this article suggests that local pension underfunding is correlated with income, the location of the county (border versus nonborder) and both in and out migration. This may be linked to other fiscal conditions that induce migration, as higher per capita property taxes are also linked to higher levels of unfunded liabilities. These findings may be broadly applicable to unfunded local pension liabilities in the United States. There are also Indiana specific considerations that play a role in unfunded liabilities. While there is no correlation to local property tax rates, income tax rates, or revenue losses from recently implemented property tax caps in either model, lower state intergovernmental revenues are correlated with higher levels of unfunded liabilities.

Retirees also play a role in unfunded pensions. Communities with higher shares of retirement income have lower levels of pension underfunding. We believe this finding to be broadly applicable but further research is needed to confirm this relationship.

We find no relationship between fiscal restrictions in the form of recently implemented property tax caps and unfunded pension liability, but recommend further research over a longer time period and other fiscal environments to assess the relationship between property tax caps and unfunded pension liabilities. These findings should inform both researchers and policymakers in their quest to better understand how unfunded liabilities emerge and are perpetuated.

Many states, including Indiana, do not require local governments to follow generally accepted accounting principles (GAAP) for government, as promulgated by GASB. Since the new GASB standards require more conservative actuarial estimates and more complete disclosure of net pension liabilities, adoption of GASB standards would ensure best practices in measuring and reporting pension liabilities. This may be a good time for more states to consider requiring use of GASB standards. Due to more conservative estimates and discount rates, local governments that apply the new standards may experience a sudden increase in the reported amount of the unfunded liability, though the underlying economics will not have immediately changed. Public officials should be prepared to communicate with their constituents about how pension liabilities are measured and about plans to deal with reported short falls. The situation is serious but not beyond hope. As researchers have noted (Munnell et al 2008, 6), most pension systems can achieve solvency if they follow "a disciplined approach to funding."

Recent changes in local pension reporting have resulted in the availability of one year of data used in this analysis. As such, this is an exploratory study. Results need to be further vetted as more data becomes available and as local governments respond to GASB rules requiring more transparent reporting of unfunded liabilities.

REFERENCES

American Academy of Actuaries. (2012) Issue brief: The 80% pension funding standard myth, Accessed April 29, 2015 (http://www.actuary.org/files/80_ Percent_Funding_IB_071912.pdf).

Chaney, B., Copley, P. and Stone, M. (2002) "The effect of fiscal stress and balanced budget requirements on the funding and measurement of state pension obligations", Journal of Accounting and Public Policy, 21(4-5):287-313.

Collins, C. & Rettenmaier, A.J. (2010) Policy Report No. 329: Unfunded liabilities of state and local government employee retirement benefit plans. National Center for Policy Analysis.

Courant, P.N., Gramlich, E.M., & Rubinfeld, D.L. (1979) "Public employee market power and the level of government spending", The American Economic Review, 69(5):806-817.

Dye, R.F. & Gordon, T.M. (2012) Pension legacy costs and local government finances, Lincoln Institute of Land Policy, Accessed March 27, 2015, (https://www.lincolninst.edu/pubs/dl/2144_1470_Pension_Legacy_Costs_and _Local_Gov_Finances_1012LL.pdf).

Epple, D., Schipper, K., & Inman, R.P. (1981) "Municipal pensions funding: A theory and some evidence", Public Choice. 37(1):141-178.

Farmer, L. (2015) "Do credit ratings matter anymore", Governing, 28(4):5457.

Farmer, L. & Maciag, M. (2015) Governing. Why some public pensions could soon look much worse, Accessed April 29, 2015 (http://www.governing.com /topics/mgmt/gov-gasb-pension-plans-may-look-worse-soon.html).

Faulk, D. 2013. The impact of property tax rate caps on local property tax revenue in Indiana, Ball State University Center for Business and Economic Research Policy Brief.

Geiger, K., Dardick, H., & Mahr, J. (2015) Governing. Illinois may be the next state to let municipalities file for bankruptcy, Accessed April 29, 2015 (http://www.governing.com/topics/finance/tns-rauner-illinoisbankruptcy.html).

Government Accountability Office (2012) State and Local Government Pension Plans: Economic Downturn Spurs Efforts to Address Costs and Sustainability. Washington: GAO.

Government Finance Officers Association (GFOA) (2012) Funding defined benefit pensions, GFOA: Chicago, IL Accessed March 27, 2015 (http://www.gfoa.org/funding-defined-benefit-pensions).

Government Finance Officers Association. (GFOA) (2015) Private electronic communication with one of the authors, GFOA: Chicago, IL.

Governmental Accounting Standards Board (GASB) (1987) Objectives of Financial Reporting. Concepts Statement No. 1, Norwalk, CT: GASB.

Governmental Accounting Standards Board. (GASB) (2012) Accounting and Financial Reporting for Pensions: An Amendment of GASB Statement No. 27. Statement No. 68. Norwalk, CT: GASB.

Indiana Gateway for local government units. (2015) Accessed February 13, 2015 (https://gateway.ifionline.org/report_builder/Default3a.aspx?rptType=&rpt=pe nsion_public&rptName=Pension).

Indiana Legislative Services Agency. (2012) Indiana Handbook of Taxes, Revenues and Appropriations. Office of Fiscal Analysis and Management Analysis,(https://iga.in.gov/static documents/4/9/2/8/49284c96/TaxHandbook12_online.pdf).

Inman, R. (1982) "Public employee pensions and the local labor market", Journal of Public Economics, 19(1):49-71.

Inman, R. (1986) "Appraising the funding status of teacher pensions: An econometric approach", National Tax Journal, 39(1):21-33. Ippolito, R.A. (1985) "The economic function of underfunded pension p lans", Journal of Law and Economics. 69(5) :611-651.

Johnson R.W. (1997) "Pension underfunding and Liberal retirement benefits among state and local government workers", National Tax Journal, 50(1): 113-142.

Kelley, D.G. (2014) "The political economy of unfunded public pension liabilities", Public Choice, 158(1):21-38.

Khumawala, S., Marlowe, J., & Neely, D.G. (2014) "Accounting & financial management", Journal of Public Budgeting, 26(2):292-312.

Lambert, L. (2013) Congress returns to public pension battle with new bill. Accessed April 29, 2015 (http://www.reuters.com/article/2013/04/18/us-usapensions-congress-idUSBRE93H0TZ20130418).

Long, S. (1997) Regression Models for Categorical and Limited Dependent Variables, Series: Advanced Quantitative Techniques in the Social Sciences, Thousand Oaks, California: Sage Publications.

Lyons, K. and Lav, I.J. (2007) The problems with property tax revenue caps, Center on Budget and Policy Priorities, Accessed April 29, 2015 (http://www.cbpp.org/files/6-21-07sfp.pdf).

Maddala, G.S. (1983) Limited-Dependent and Qualitative Variables in Econometrics, Cambridge: Cambridge University Press.

Marks, B.R. & Raman, K.K. (1996) "The behavior of interperiod equity-related performance measures over time", Accounting Horizons, 10(4):52-66.

Marks, B., Raman, K. and Wilson, E. (1988) "Toward understanding the determinants of pension underfunding in the public sector," Journal of Accounting and Public Policy, 7(3):157-183.

Moodys Investor Services (2012) US State and Local Governments Face Risks with Pension Funding Bonds.

Moodys Investors Services (2013) Adjusted Pension Liability Medians for US States.

Mortimer, J.W. & Henderson, L.R. (2014) "Measuring pension liabilities under GASB Statement No. 68", Accounting Horizons, 28(3):421-454.

Munnell, A.H. (2012) Myths and Realities about State and Local Pensions from State and Local Pensions: What Now? Center for Retirement Research. Accessed online February 22, 2015 (http://crr.bc.edu/wpcontent/uploads/2012/09/myths_realities_appd.pdf).

Munnell, A.H., Aubry, J.P., & Cafarelli, M. (2015) How Did State/Local Plans Become Underfunded? Boston: Center for Retirement Research, Boston College.

Munnell, A.H., Haverstick, K., Sass, S.A., & Aubry, J.P. (2008) The Miracle of Funding by State and Local Pension Plans, Boston: Center for Retirement Research, Boston College.

Novy-Marx, R., and Rauh, J. (2011) The crisis in local government pensions in the United States. In Growing old: Paying for retirement and institutional money management after the financial crisis, Robert Litan and Richard Herring, eds., 47-74. Washington, DC: Brookings Institution.

Pew Charitable Trusts (2012) Widening gap update: Indiana, Accessed March 25, 2015 (http://www.pewtrusts.org/en/research-and-analysis/fact-sheets/2012/ 06/18/widening-gap-update-indiana).

Rauh, J. (2011) NBER Reporter 2011 Number 1: Research summary. The impact of employee pension promises on state and local public finance, Accessed February 27, 2015 (http://www.nber.org/reporter/2011number1/rauh.html).

Rich, K. & Zhang, J. (2014) "Unfunded public pension liabilities and local citizen oversight", Accounting Horizons, 29(1):23-39.

United States Congress, House Committee on Education and Labor (1978) Pension Task Force Report on Public Employee Retirement Systems. Washington, DC.

Vermeer, T., Styles, A. and Patton, T. (2010) "Are local governments adopting optimistic actuarial methods and assumptions for defined benefit pension plans?" Journal of Public Budgeting, Accounting & Financial Management, 22(4):511-542.

Vermeer, T., Styles, A. and Patton, T. (2012) "Do local governments present required disclosures for defined benefit pension plans?" Journal of Accounting and Public Policy, 31(1):44-68.

White, Halbert. (1980) "A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity", Econometrica: Journal of the Econometric Society, 48(4):817-838.

Dagney Faulk, Ph.D. (Corresponding author)

Director of Research

Center for Business and Economic Research

Ball State University

Muncie, IN 47306

(765) 285-5152

dgfaulk@bsu.edu

Michael Hicks, Ph.D.

Director and Professor

Center for Business and Economic Research

Ball State University

Muncie, IN 47306

mhicks@bsu.edu

Larita Killian, Ed.D.

Associate Professor of Accounting

Indiana University at Columbus

Columbus, IN

812-348-7219

ljkillia@iupuc.edu

(1.) GASB also issued new standards for pension plan administrators (such as fiduciary trusts), Statement No. 67: Financial Reporting for Pension Plans. GASB Statement No. 68 provides guidance for local governments who sponsor plans.

(2.) The "80%" benchmark was cited by Inman (1986, 27) and several writers since. The American Academy of Actuaries (2012, 2-3) and the GFOA (2012) argue against use of a single benchmark.

(3.) The new GASB standard (2012) allows variation in discount rates, but a more conservative composite rate will be required for calculating projected benefit obligations.

(4.) The Public Pensions Database is maintained by the Center for State and Local Government Excellence and the Center for Retirement Research at Boston College.

(5.) Local governments in 76 of Indiana's 92 counties have defined benefit plans. See figure 1. LaPorte County local governments have defined benefits plans, but the county does not provide data on the property tax variables used in the model and is therefore not included in the analysis.

(6.) School district pension plans are not included in this analysis.

(7.) See Madalla (1983) for more information about and the functional form of the extreme value distribution.

(8.) Ideally, we would like to include a variable measuring the percentage of local government workers who are members of labor unions, but this data is not available at the county level. The majority of the pension plans included in this analysis is for law enforcement (police and sheriff) which are highly unionized.

(9.) Additional model specifications were tested and are available from the authors upon request.
Table 1. Explanatory variables

Dependent Variables              State of      Fiscal       Interest
                                 the Economy   Structure/   Group
                                               Stress       Variables

Per capita income ($)            X

Unemployment rate (%)            X

MSA dummy                        X

Border dummy                     X

Percent change in county         X
population 2003-2013 (%)

In migration 2008-2012 as a      X
share of the county population
(%)

Outmigration 2008-2012 as a      X
share of the county population
(%)

Net property tax levy per                      X
capita ($)

Effective (average) property                   X
tax rate (%)

Property tax cap credit per                    X
capita ($)

Local option income tax rate                   X
(%)

State intergovernmental                        X
revenue per capita ($)

Retirement income per capita                                X
($)

Local government employment as                              X
a share of total county
employment (%)

Current local government                                    X
retirees as a share of local
government employment (%)

Vested local government                                     X
employees as a share of local
government employment (%)

Dependent Variables              Source

Per capita income ($)            Census SAIPE

Unemployment rate (%)            BLS

MSA dummy                        OMB

Border dummy

Percent change in county         BEA
population 2003-2013 (%)

In migration 2008-2012 as a      ACS
share of the county population
(%)

Outmigration 2008-2012 as a      ACS
share of the county population
(%)

Net property tax levy per        DLGF
capita ($)

Effective (average) property     DLGF
tax rate (%)

Property tax cap credit per      DLGF
capita ($)

Local option income tax rate     INH
(%)

State intergovernmental          COG
revenue per capita ($)

Retirement income per capita     BEA
($)

Local government employment as   COG, BEA
a share of total county
employment (%)

Current local government         Indiana Gateway
retirees as a share of local
government employment (%)

Vested local government          Indiana Gateway
employees as a share of local
government employment (%)

Sources: ACS (American Community Survey), BEA (Bureau of Economic
Analysis, US Department of Commerce), BLS (Bureau of Labor
Statistics), INH (Indiana Handbook of Taxes, Revenues and
Appropriations), COG (2007 Census of Governments), DLGF (Indiana
Department of Local Government Finance OMB (U.S. Office of
Management and Budget), Indiana Gateway for government units.

Table 2. Descriptive statistics

                                           Mean     Median   Std Dev

Unfunded local government pension          28.42    19.82    40.46
liability per capita ($)

Unfunded local government pension          2,467    1,640    4,424
liability per local government employee
($)

Per capita income ($)                      37,214   36,956   4,969

Unemployment rate (%)                      7.47     7.30     1.16

MSA dummy                                  0.49     0        0.50

Border dummy                               0.40     0        0.49

Percent change in county population        2.68     1.11     7.85
2003-2013 (%)

In migration 2008-2012 as a share of the   6.26     5.40     6.19
county population (%)

Outmigration 2008-2012 as a share of the   6.58     5.79     6.99
county population (%)

Net property tax levy per capita ($)       863.70   854.38   227.61

Effective (average) property tax rate      2.03     2.04     0.67
(%)

Property tax cap credit per capita ($)     60.75    36.22    72.75

Local option income tax rate (%)           2.97     2.76     1.73

State intergovernmental revenue per        1,261    1,195    326
capita ($)

Retirement income per capita ($)           3,164    3,192    428

Local government employment as a share     2.99     2.61     1.45
of total county employment (%)

Current local government retirees as a     1.05     0.65     1.50
share of local government employment (%)

Vested local government employees as a     2.89     1.73     4.77
share of local government employment (%)

                                           Min      Max

Unfunded local government pension          0        329.72
liability per capita ($)

Unfunded local government pension          0        37,410
liability per local government employee
($)

Per capita income ($)                      27,627   56,515

Unemployment rate (%)                      5.30     10.60

MSA dummy                                  0        1

Border dummy                               0        1

Percent change in county population        -8.51    37.90
2003-2013 (%)

In migration 2008-2012 as a share of the   0.99     56.05
county population (%)

Outmigration 2008-2012 as a share of the   1.03     64.70
county population (%)

Net property tax levy per capita ($)       1.31     1,445.20

Effective (average) property tax rate      0        3.79
(%)

Property tax cap credit per capita ($)     0.04     357.54

Local option income tax rate (%)           0.00     7.66

State intergovernmental revenue per        689      2,275
capita ($)

Retirement income per capita ($)           2,056    4,199

Local government employment as a share     1.11     7.53
of total county employment (%)

Current local government retirees as a     0        8.62
share of local government employment (%)

Vested local government employees as a     0.47     30.84
share of local government employment (%)

Note: The number of observations is 75.

Table 3. Estimation results

                             Unfunded liability   Unfunded liability
                             per capita           per local government
                                                  employee

                             Coefficient          Coefficient
                             [p-value]            [p-value]

Constant                     498.55 ***           57,166 ***
                             [0.0003]             [0.0002]

Per capita income ($)        -0.005 **            -0.623 ***
                             [0.0137]             [0.0092]

Unemployment rate (%)        7.050                854.116
                             [0.4531]             [0.4105]

MSA dummy                    22.528               2705.339
                             [0.1381]             [0.1072]

Border dummy                 -27.554 **           -3,095.06 **
                             [0.0371]             [0.0356]

Percent change in county     -0.715               -82.861
population 2003- 2013 (%)    [0.6453]             [0.6315]

In migration 2008-2012 as    9.231 *              1,166.37 **
a share of the county        [0.0741]             [0.0410]
population (%)

Outmigration 2008-2012 as    -9.1614 **           -1,098.32 **
a share of the county        [0.0326]             [0.0204]
population (%)

Net property tax levy per    0.0883 **            9.1574 **
capita ($)                   [0.0235]             [0.0342]

Effective (average)          -18.935              -2,078.75
property tax rate (%)        [0.3152]             [0.3253]

Property tax cap credit      -0.155               -21.191
per capita ($)               [0.2997]             [0.2077]

Local option income tax      6.759                546.667
rate (%)                     [0.1070]             [0.2353]

State intergovernmental      -0.0008 ***          -0.0899 ***
revenue per capita ($)       [0.0003]             [0.0002]

Retirement income per        -0.0809 ***          -8.6716 ***
capita ($)                   [0.0002]             [0.0003]

Local government             0.767                -240.001
employment as a share of     [0.8596]             [0.6138]
total county employment
(%)

Current local government     2.644                192.828
retirees as a share of       [0.2019]             [0.4184]
local government
employment (%)

Vested local government      -0.457               -78.6504
employees as a share of      [0.5410]             [0.3697]
local government
employment (%)

Obs.                         75                   75

Akaike information           10.755               19.785
criterion

*** p < 0.01 ** p < 0.05 * p < 0.10
COPYRIGHT 2016 Southern Public Administration Education Foundation, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2016 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Faulk, Dagney; Hicks, Michael; Killian, Larita
Publication:Public Finance and Management
Article Type:Report
Geographic Code:1U3IN
Date:Mar 22, 2016
Words:8587
Previous Article:The impact of late-term pay raises on teacher pension obligations in three states: California, Illinois, New Jersey.
Next Article:Rural-urban migrants' citizenization: public services cost measurement and economic impact analysis.
Topics:

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters