Background on Price Index Construction

Illinois Municipal Price Index Revisions and Updates Norman Walzer, Andy Blanke, and Michael Aquino 1 Managing local public resources is especially challenging during periods of slow economic growth and recessions which make balancing budgets difficult. Comparisons of expenditures or revenues over time are often essential parts of this process and require a detailed understanding of the impacts of past decisions about resource reallocations. Since municipal services involve mainly personal expenditures with the outputs provided being difficult to measure precisely even though increasingly municipalities track performance measures and outcome indicators. Thus, local decision-makers often rely heavily on expenditure comparisons to estimate resources available for local services. These comparisons within a municipality, or among municipalities, are affected by price changes as well as variations among locations. While detailed data on differences in prices of items purchased by various municipalities are not readily available and therefore are not included in this project, the Illinois Municipal Price Index (MPI) provides a way to compare the impact of price increases over time. This report describes the approaches used to revise the MPI as well as update it for 2016. It also uses the MPI to describe changes in municipal expenditures and revenues during the 2009 recession and the recovery. Background on Price Index Construction Inflation is estimated by taking the percentage distribution of expenditures by category for a specific department and then applying it to the price increase for each respective category. Afterward, the products for each of these expenditures are summed and divided by the sum of those in the previous year. Given that a large portion of municipal expenses are for personnel, municipalities are more affected by inflation than the private sector (Walzer & Davis, 2015). The MPI is preferred for measuring inflation compared with using the Consumer Price Index (CPI) or Producer Price Index (PPI) because it considers a mix of purchases unique to municipalities and not purchased by private businesses or individual consumers. The MPI also involves examining price changes for each department in a municipality since inflation can affect specific departments differently depending on composition of expenditures. 1 The authors are Senior Research Scholar, Research Assistant, Economics Graduate Student in the Center for Governmental Studies at Northern Illinois University. They thank the Illinois Municipal League staff for assistance in data collection and financial support of the project. 1

MPI vs CPI Prices are typically calculated in one of two ways: Paasche indices and the Laspeyres indices and these methods differ mainly in the base weights used. A Laspeyres Index uses a fixed base of quantities while Paasche indices have variable weights. The Paasche index tends to understate inflation while the Laspeyres index may overstate inflation since the weights do not include adjustments due to price increases through time. The MPI was originally constructed in 1969 and the construction approach has remained similar since then although the weights have been updated at various times (Walzer, 1970). The CPI has been studied extensively in terms of construction and appropriate use because of its applicability to various scenarios. In addition, there has been considerable research on problems that arise when using the CPI to account for inflation as has been addressed by Boskin (2005) and Nakamura et al. (2014). Boskin (2005) reviewed the causes of upward bias in the CPI found in some Bureau of Labor Statistics (BLS) estimations in the past and claims that when taking the CPI as a measure of true cost of living index (COLI), the CPI overstates inflation by approximately one percent annually. This implies an upper level substitution bias, i.e. the CPI did not include the effects of consumers changed purchases from one category to another due to price changes. When calculating the CPI, the BLS creates the fixed-weight Laspeyres Index using base period expenditure weights obtained from consumer expenditure surveys (Boskin, 2005). Nakamura et al. (2014) assert that for the CPI, issues arise when there is a clear and present promotional or outlet substitution bias. This bias occurs from failing to capture changes in purchasing due to temporary sales and other prices differences among retailers (Nakamura et al., 2014). In addition, because the CPI calculates price changes from month to month, it does not account for how consumers spend during limited, e.g. weekend-only, sales. Consequently, the CPI can overstate or understate inflation depending on the prevalence of promotions for a specific good or service. Compared with the CPI, the MPI is less sensitive to distortion due to temporary retail promotions. Many purchases by municipalities are contractual, and so prices are more likely to remain stable from month to month. The MPI estimates inflation specific to municipalities better than the CPI primarily because it omits the possible sales bias found in the CPI. Also, the MPI heavily weights changes in wages and salaries which are usually annual contracts. MPI vs PPI Inflation is estimated in the PPI similarly to the CPI, but the PPI specifically measures inflation in business-to-business purchases and business sectors other than retail (Walzer & Davis, 2015). Traditionally, the PPI uses the Laspeyres method of calculation and so the problems of CPI also apply to the PPI (Park and Ryu, 2011). Overall, the use of the PPI can help categorize specific goods and services and identify complimentary goods/services. Norman (2008) notes though, that a particular disadvantage of the PPI is that the actual values of prices and price relatives for these production goods are not clear. In short, it is not possible to tell from PPI data actual prices/price relatives, only 2

relative/percentage movements in prices and price ratios over time. This is due to the data being hard to retrieve as most firms are unwilling to share purchase information because it could benefit competitors. Another issue with PPI estimation methods involves sourcing substitution bias. Nakamura et. al. (2014) state that substitution bias occurs when businesses purchase their inputs from different sources due to changes in prices. For example, an automotive parts manufacturer may purchase sheet metal from a new supplier that can offer the metal at a lower price. As described for the CPI, if businesses substitute input purchases during price inflation, the PPI can overestimate inflation. The MPI by comparison is less affected by these issues due to the nature of purchases. Municipal purchases are often contractual over longer terms, meaning that cities change suppliers less frequently depending on projects. Therefore, as for the CPI, the MPI offes more reliable inflation estimates for municipal purchases. Price Index Innovations Since both CPI and PPI have estimation bias issues, researchers have found ways to improve the accuracy of inflation estimates and price indices are being constructed using more complex methodologies. As mentioned, Nakamura et al. (2014) and Boskin (2005) cite CPI biased estimation problems but also offer solutions. Nakamura et al. (2014) recommended an alternative to conventional price indexes including information on quantity of goods purchased, in order to combine transactions that occur at different prices for the same product units. The approach would be to collect data on both the value of total sales in a particular month and the total quantities sold in order to derive a unit value to use as the price (Nakamura et al., 2014). The MPI uses a similar methodology (Walzer, 1970). Both methods take into account the unit share of the sales/expenditures for a specific product or service. The main difference is that the MPI includes a percentage distribution of expenditures as opposed to Nakamura et al. s (2014) unit values. As for the PPI, chained-type estimation of the price index is one way to reduce estimation bias. A chained-type price index resets the base year each year which helps to mitigate the tendency to overestimate inflation. Park and Ryu (2011) offer the chained price index methodology as a possible way to correct the inherent upward bias in the PPI. However, there are a couple of drawbacks in estimating the PPI using this approach: Ideally, weights for the previous year would be used in calculating the chained-type index, but this is usually not possible due to time lags in collecting data for those weights. (Park and Ryu, 2011) Between the months of December and January the following year, there is usually a jump in the series of chain-type price indices due the difference of weights used, since one month is at the end of the year and the other month is at the beginning of the next year. (Park and Ryu, 2011) 3

By switching from the Laspeyres index to the chained-type index, the inherent statistical property of the index may also change. Because of this, misspecification could be an issue. (Park and Ryu, 2011) Despite these drawbacks, Park and Ryu, (2011) find that there is a high level of substitutability between fixed-based indices and chain-type indices. In short, Park and Ryu, (2011) find that the chain-type index provides more statistical consistency and stability over a fixed-based index. The MPI uses fixed base year weights which are updated approximately every 10 years and it would not be feasible to revise the weights on an annual basis. In summary, inflation remains a difficult concept to measure in the public and private sectors, and improvements are underway in the MPI, PPI, and CPI. Price Index Structure The intent of a price index is the determine the impact on local revenues or expenditures from price changes during a specified period. The two main components of the MPI are the price changes of specific groups of goods and services purchased and the relative importance of each of these categories. While there are several ways to construct an index, the most straightforward and the one used in past MPI constructions is the weighted aggregate of price relative index where the price changes are weighted by the percentage that each purchase category represents in either a department or the aggregate municipal budget. It is important to have accurate expenditure information on the specific user group for which the price index is constructed. The Consumer Price Index, for example, is based on purchases by households and related individuals. Thus, the CPI includes price changes for housing, food, transportation and other items in the relative importance they have for an annual budget. Other indices such as the Producer Price Index have different sets of weights. This means municipal officials and administrators may not find the CPI or the PPI of much direct use in monitoring the resource allocations among municipal services since the purchase categories and relative ways are different. The MPI is especially useful in Illinois because changes in wages and salaries, the largest component of municipal expenditures, are based on experiences in an Illinois sample that will be described later. For comparison purposes, Table 1 provides information on several indices that quantify the effects of inflation on both private and public groups. The Consumer Price Index (CPI) is the most often cited impact of inflation and is based on a market basket of goods and services purchased by families and unrelated individuals. The CPI includes food, housing, transportation, entertainment, and similar types of purchases which makes it less relevant for municipalities. 4

Since three-quarters or more of the purchases by municipalities are for personnel, the Services component of the CPI can help understand movements in the MPI and are discussed here. In the past year, the price changes for Commodities (-1.40%) differ markedly from those of Services (2.81%) suggesting that, on average, private sector workers probably gained slightly in purchasing power. To compete for labor with the private sector, public agencies must pay competitive wages, so the municipal compensation packages must increase at approximately the same rate as in the private sector. This is especially true in recent years with the overall unemployment rate approaching what has traditionally been considered full-employment level. The total MPI increased 2.81% during 2016 reaching a value of 157.3 and indicating that it required $157.30 in 2016 to purchase what $100.00 would have bought in 2000. The higher MPI value most likely reflects the higher concentration of personal services in the MPI than in the CPI but is closer, although smaller, than the 2.81% for Services in the CPI. The American City and County Municipal Cost Index (ACCMCI) (http://americancityandcounty.com/municipal-cost-index) started in 1978 and uses national data on municipal expenditure categories to estimate changes in the costs of providing municipal services. The MPI differs from the ACCMCI in two ways. First, the MPI is constructed using more detailed spending information for municipalities. Second, the MPI is based, as much as possible, on data for Illinois municipalities, especially wage and salary increases and pensions discussed below. Nevertheless, a comparison between the indices sheds light on the differential impact of price increases on Illinois municipalities compared with a national sample. On average, the MPI and ACCMCI track fairly closely with the MPI running slightly ahead until 2015 when the ACCMCI showed a decrease. The ACCMCI contains less detail on spending patterns and does not show Illinois experiences especially regarding personnel costs and pensions. If Illinois spending patterns or personnel costs run ahead of the national experience, the MPI will be higher. Nevertheless, the MPI exceeds the CPI Services component as well as the ACCMCI confirming past experiences where local employees gain in purchasing power during periods of low inflation in the private sector. 5

Table 1. Price Indices (2000=100.0) Pct. Chg. Index 2000 2005 2010 2012 2013 2014 2015 2016 2015-16 Consumer Price Index All Items 100.0 113.4 126.6 133.3 135.3 137.5 137.6 139.4 1.31 Commodities 100.0 107.4 117.0 125.7 125.8 126.0 121.8 120.1-1.40 Services 100.0 117.8 133.8 139.0 142.3 146.0 149.4 153.6 2.81 Producer Price Index 100.0 118.6 139.2 152.4 153.3 154.7 143.6 139.7-2.72 American City & County Municipal Cost Index (January, 2000 = 100) 100.0 116.7 135.8 145.8 148.1 150.2 152.1 150.8-0.87 Illinois Municipal Price Index 100.0 118.4 137.8 144.1 147.4 150.6 153.0 157.3 2.81 NOTE: The entire series is available at http://www.iira.org/data/mpi.asp Source: U.S. Department of Labor, Bureau of Labor Statistics, American City and County, and Center for Governmental Studies at Northern Illinois University. 6

Expenditure Weights Expenditure weights in the MPI were determined from the review of detailed expenditure reports from 20 Illinois municipalities of various sizes and locations in Illinois including representation by Chicago suburbs, downstate metropolitan areas, and rural areas. In some cases, the standardized annual financial reports that Illinois municipalities submitted to the Illinois Office of the Comptroller did not have sufficient detail to match expenditure categories in the price index. In these instances, municipal officials and administrators provided a treasurer s report or a statement of line item expenditures. Expenditure reports that were provided were reviewed individually with items sometimes reclassified into the 40-50 expenditure weight categories in each municipal department. This information was typically reported by department, with categories that easily fit the price index groupings of purchases. If not, expenditures were reported by fund, with the General Fund including expenditures for multiple department areas, such as general control, planning/development, and public works. In these cases, general fund expenditures were separated into component departments as appropriate. For example, if the salary for an economic development director/coordinator was listed as a general fund expenditure, then it was reclassified as an expenditure for an economic development supervisor rather than a general control supervisor. Another example involved engineer salaries listed in the general fund were classified under public works rather than general control. Once expenditure data from all received reports had been reclassified into the MPI expenditure categories, expenditure weights were calculated as the weighted average percentage of city and departmental expenditures. Below are example formulas illustrating how departmental and citywide weights were calculated for police pensions: Department Weight, Police Pensions = Citywide Weight, Police Pensions = Total Police Pension Expenditures Total Police Department Expenditures Total Police Pension Expenditures Total City Expenditures, All Departments Inconsistencies in reporting compensation by employee type arose in converting expenditure data to expenditure weights. Most municipalities consistently reported compensation levels for full-time employees compared to temporary or part-time employees, but not all distinguished between labor, clerical, and supervisory positions. Given the weighting scheme in the previous price index, this led to an increase in weight for skilled/professional labor in several departments, and a decrease in weights for supervision and clerical costs. Weights from the current review of 7

expenditures were used retroactively to recalculate price index values for previous years to prevent sudden changes in index values due to differences in the sample of 2016 expenditures compared with previous years. Following is a discussion of changes in expenditure weights for the city and departmental indices. 2 In general, the departmental weights changed more than the city weights because changes in expenditures for one department could be offset by trends in other departments in the aggregate index. The departmental indices reflect expenditure changes within the same departments, while the city-wide index reflects changes across the sample cities, including those without some departments, such as an independent library district or a volunteer fire department. General Control Most city and departmental weights remained within 1% of their values in the previous model. However, there was an increase in skilled/professional labor and a decrease in expenditures for supervision and clerical work, possibly due to differences in the level of payroll data available. Also, the weights for pensions increased 1.7% in the departmental index compared with last year, meaning that pensions represented a larger share of total general fund expenditures. In summary, the expenditure weights for General Control were relatively minor, meaning that expenditures from central departments such as the manager/administrator s office and the finance department involved essentially the same mix of goods and services as in previous years. Community and Economic Development For the city-wide index, all weights remained within 0.5% of those in the previous MPI. However, a few significant changes were made to the departmental weights based on the expenditure summaries. The weights for labor expenditures decreased 7.3% while the weights for supervision decreased 7.9%. Some municipalities listed only the salary of an economic development director or coordinator and other cases made no distinction in payroll for staff compared to supervisors. These differences could explain the similar trend in weights for both types of personnel. There also was a 3.8% increase in capital expenditures for computers/office machinery and a 7.5% increase for vehicle purchases. Community and economic development departments in the sample seem less labor intensive and more capital intensive than in previous MPIs and the specific reasons are not clear. 2 In these discussions the city index refers to the aggregate MPI for the municipal expenditures as a whole and the departmental index refers to each municipal department. 8

Other Sanitation (Excluding Sewer) Most weights in the city index remained within 0.5% of those in the previous MPIs. One significant change was a 1.3% increase in labor, which was accompanied by minor decreases in the share of expenditures for contractual services. This trend was more extreme in the departmental index, where labor increased 41.2% and contracts for maintenance/repair and special services each decreased more than 5%. The cities used to construct the current weights may employ more repair and sanitation workers in-house compared to the previous MPI, when cities may have contracted with private businesses for more of these services. Relatively few of the municipalities providing expenditure data reported expenditures regarding sanitation separate from sewers, so the comparability of these weights may be affected by data differences. However, the MPI is calculated based on many types of purchases, so changes in one variable are mitigated by others, e.g. a weight that changed significantly could be applied to a price relative that remained relatively stable and the overall effect was minimal. Health All weights for the city index remained within 0.5% of the previous weights. In the departmental index, the share of expenditures in personal services decreased slightly while special services contracts increased. There was also a 2.7% increase in capital expenditures for computers and machinery. Water & Sewer Most weights in the departmental index remained within 0.5% of those previously used but with a 1.6% decrease in capital expenditures for pipe maintenance and a 1.5% decrease for other contractual services. The departmental index had a 1.7% decrease in the weight for labor that was offset by increases for utilities and land/building water improvement. The decreases in pipe maintenance may be due to expenditures classified as water system maintenance in the expenditure reports. Parks and Recreation All weights in the city index remained within 0.3% of the previous weights. Most weights were also unchanged in the departmental index, but there was a 2.9% increase in skilled/professional labor and a 3.4% decrease in temporary help, most likely due to differences in expenditure report formats. When payroll expenditures were listed without specifications of employee status or responsibilities, they were coded by default as skilled/professional labor. There was also a 7% increase in capital expenditures for land/building improvements and a 3.7% increase in vehicle purchases. Relatively few municipalities providing expenditure data reported expenditures from a parks/recreation department, so large capital projects in one or two cities could represent a larger share of total park district expenditures compared with a previous year. Care was taken to remove large capital projects from the expenditures when possible. 9

Police The city index weights remained within 0.4% of the previous MPI. In the departmental index, most weights also were relatively unchanged, but with a 10.3% decrease in labor offset by a 10.8% increase in pensions and a 2.8% increase in temporary help. Some cities may have experienced retirements with hires of new officers at lower salaries but are now making higher pension or OPEB payments. Fire Most weights in the city and departmental indices were relatively unchanged. Exceptions include an increase in pensions (2.7% city and 10.7% departmental) and a decrease in telephone/communications contracts (-1.2% citywide and -7.5% departmental). Labor decreased 6.7% in the departmental index, but its weight remained within 1% of the previous city weight. As in police departments, some cities may have had experienced employees retire. The decreased communication expenditures may mean that technology improvements and shared services reduced costs in responding to emergency calls. Streets In the city index, the weights decreased slightly in some labor and contractual expenditures. Skilled/professional labor decreased 1.6%; engineering/property maintenance contracts decreased 1.4%; and sidewalk maintenance contracts decreased 1.2%. However, special services contracts increased 1.1%. The weights for supplies and capital expenditures remained within 0.5% of the previous city MPI. Weights for the departmental streets index followed similar trends. Skilled/professional labor decreased 2.4% while property maintenance contracts decreased 6.9% and street maintenance contracts decreased 2.8%. Some of these decreases were offset by an 8.2% increase in special services. Capital investment increased significantly from the previous year, including an 8.6% increase in the departmental weight for building/land/streets expenditures and a 4.1% increase in vehicle purchases. Library All weights remained within 0.3% of the previous MPI in the city index. In the departmental index, skilled/professional labor increased 11.1% while weights in the other employee classes decreased. Relatively few cities in the sample reported library expenditures and, in the data provided, most cities did not differentiate payroll expenditures by employee type, e.g. supervision vs clerical. Also, the weight for special services contracts increased 10.0% while those for other contracts decreased, due to lesser detail reported in the contractual services sections of the expenditure reports. 10

Of special interest is that the weight for book purchases decreased 6.8% and the weight for audio/visual media increased 1.3%, suggesting that the libraries examined are transitioning toward digital media. Finally, the weight for capital expenditures on building improvements increased 15.6%. Because relatively few sample municipalities provided expenditure information on libraries, large capital projects in one municipality could represent a larger share of total expenditures. Price Relatives Price relatives were collected from an annual compensation survey conducted by the Illinois Municipal League and several national data sources described below. Selected items from the detailed Consumer Price Index published by the Bureau of Labor Statistics are used to measure price changes for specific services and commodities. The CPI, All Commodities decreased -2.5% between 2015 and 2016, mainly due to lower prices of petroleum products. The CPI, Services increased mainly because of rising prices for health care, health insurance, and real estate, among other items. The BLS also publishes a Producer Price Index, used to estimate price changes for commodities and other items purchased in large quantities. The overall PPI decreased between 2015 and 2016, due mainly to motor fuels (9.0% decrease), processed foods, and metals. Since some municipal departments purchase large quantities of items such as petroleum products and concrete, price changes on specific items are obtained from the PPI. The BLS also publishes an Employment Cost Index, based on a survey of wages and benefits at 6,800 private companies and 1,400 government agencies. This index increased 3.1% during the previous year, reflecting both health insurance costs and compensation increases. The Employment Cost Index is used to estimate changes in pension expenses. Attempts were made to include information from Public Pension Reports published by the Illinois Department of Insurance, but this data is only available biennially and 2016 data was unavailable at the time of revising the MPI. Additional efforts will be made to incorporate changes in pension costs in future years. Another source for price relatives is a building cost index published by Turner Construction Company. This index measures changes in nonresidential construction costs, based on labor costs, productivity, material costs, and the level of competitiveness in the construction market. This index increased 4.9% between 2015 and 2016. Price changes for investments are obtained from Federal Reserve data that includes bond yields and interest rates on mortgages. In previous years, the interest rate on corporate bonds 11

rated BAA by Moody s had been used to estimate the typical cost of investments, but this data was discontinued in 2016. For this reason, the effective yield on bonds rated BBB by Bank of America Merrill Lynch, available for 1997 through the present, was used. It follows a trend similar to the discontinued BAA dataset. In part because of this change in source data, MPI values for previous years were retroactively updated so differ slightly from past publications. Price relatives for employee compensation are obtained from an annual compensation survey by IML. In total, 26 cities responded to the 2016 survey, although not all provided complete information for each occupation. 3 Only cities with data for both 2015 and 2016 were used in the calculations to remove sample distortions. Also, cases of a drastic change between years, e.g. decreases due to hiring a less experienced replacement because of a retirement were excluded. Those not in the survey or which had not responded in 2015 were asked to provide information in that year along with the 2016 compensation levels. Compensation increased for all positions in the survey relative to 2015, for both labor and supervisory positions. Salaries rose at the same rate for police chiefs and patrol officers. Firefighter salaries grew slightly faster than fire chief salaries, but this may be affected by the fewer respondents providing usable data for this position. The limited number of municipalities with compensation information for park supervisors or library administrators can be explained by park or library districts serving those areas. How have Municipal Finances Changed? As noted previously, inflation and economic difficulties can change municipal finance patterns, the impact of which can sometimes be seen by examining the changes in constant dollars after several years. For instance, during a major recession and shortages of revenues, capital projects can be reduced or delayed to preserve current services. This section examines municipal finance patterns between 2007 to 2014 to determine how both expenditures and revenues have changed in per capita amounts (constant dollars) as well as their relative importance in the budget. These comparisons are provided in Table 2. The revenue and expenditure trends are examined using the five-year Census of Government Finances along with the latest available data from the annual Survey of Government Finances also published by the Census of Governments. All Illinois municipalities complete the five-year survey in years ending in 2 and 7 but only 161 Illinois municipalities completed the survey in 2014. For this year, all municipalities with populations between 25,000 and 50,000, (25,000 being the minimum population for home rule status and 50,000 being the minimum population 3 See Appendix for a full list of cities in the survey. 12

for designation by the Office of Management and Budget as the center of a metropolitan area) were examined yielding a sample of 51 cities. Weighted average revenues and expenditures were examined for this sample using 2007 and 2014 data using the MPI to adjust for inflation. Table 2. Revenues and Expenditures Per Capita (Mid-Size Illinois Municipalities*, 2007-2014) Dollars Per Capita % of Total % Change, 2007-2014 Category 2007 2014 (Current) 2014 (Constant) 2007 2014 Current Constant Total Revenue $1,448 1,795 1,532 24.0% 5.8% Property Tax 285 386 330 19.7% 21.5% 35.3% 15.5% Charges For Service 165 223 190 11.4% 12.4% 35.1% 15.3% From State 354 409 349 24.5% 22.8% 15.3% -1.6% From Federal 14 14 12 1.0% 0.8% -1.5% -16.0% Total Expenditures $1,414 1,584 1,351 12.0% -4.4% Current Operations 1,037 1,370 1,169 73.3% 86.5% 32.2% 12.8% Police 257 324 277 18.2% 20.5% 26.0% 7.5% Fire 151 188 161 10.7% 11.9% 24.5% 6.3% Highways 187 207 177 13.2% 13.1% 10.6% -5.6% Capital Outlays 266 210 179 18.8% 13.3% -21.0% -32.6% *51 municipalities with populations between 25,000 and 50,000. Percentages do not total 100 because not all components are listed. Source: U.S. Census Bureau, 2007 Census of Government Finances & 2014 Annual Survey of Government Finances. Inflation adjustments were calculated from the Illinois Municipal Price Index. Revenues Current revenues increased from $1,448 to $1,795 (24%) per capita between 2007 and 2017 for the same group of respondents. While that change may initially seem larger than initially expected, when price increases are removed, the increase was 5.8% or less than 1% per year. Likewise, annual population counts are not available for respondents so population declines in this period may slightly overstate the per capita averages. Some shifts in types of revenues collected also occurred during this period. For instance, property taxes collected went from $285 to $336 per person (35.3%) in current dollars but when inflation is removed, the increase was 15.5%, again recognizing that population changes affect these numbers. The fact that property taxes increased during this period from 19.2% of total 13

revenues to 21.5% may well reflect declines in other revenue source such as aid from the federal or state government. While aggregate state aid per capita to these municipalities increased 15.3% in current dollars during this period but, in constant dollars, decreased 1.6%. As a percentage of total revenues, property taxes went from 24.5% to 22.5%. The state aid figures have some distortion because detailed data on types of state aid are not available for all 51 sample municipalities. It may be in a few instances that a capital grant is included in the figures which could affect the comparisons. However, overall, the relationships are as expected. Municipalities receive relative little revenue directly from the Federal government unless they have grants for specific projects such as HUD, transportation, or from another agency. However, the trend in Federal revenues is similar to that for the state with a decrease of 16.0% in per capita, causing the importance of this revenue source to decrease from 1.0% to 0.8% during this period. Another revenue source available to municipalities during tight fiscal times involves charging for services. Table 1 has two types--charges for utilities and other charges. We recognize that charges for utilities are often handled separately from the general fund so are not available for regular operating costs as would be true for the general fund. Charges other than utilities collected by municipalities increased from $165 per person to $190 per person (35.1%) during this period and 15.3% in constant dollars. Nevertheless, they had about the same relative importance in both 2007 and 2014 (11.4% and 12.4%). This trend suggests that during a period of tight finances, municipalities did not significantly add user charges for services to make up revenue deficits. Expenditures Municipal expenditures per capita decreased (in constant dollars) from $1,414 to $1,351 (4.4%) between 2007 and 2014. The decrease includes deferred capital expenditures and higher operating expenditures. During tight fiscal periods, local officials may be pressured to delay or reduce capital projects to continue current operations and that seems to have happened. The average city spent $266 per resident on capital outlays in 2007 compared with $179 in 2014 (a 32.6% decline). More specifically, highway expenditures decreased 5.6% while police expenditures increased 7.5% and fire protection expenditures increased 6.3%. Both represented a larger share of expenditures compared with 2014. The share of municipal expenditures represented by police increased from 18.2% to 20.5% and the share of expenditures for fire increased from 10.7% to 11.9%. 14

The trends described above are in general terms and a more detailed analysis of postrecession municipal revenue and expenditure trends using a larger sample of cities, will be possible as more current data is released in the 2017 Census of Government Finances in 2019. Nevertheless, this brief discussion suggests that decided shifts in both revenues and expenditures occurred during the recession and the subsequent recovery. The changes confirm past strategies. Summary Local municipal officials and managers have used the MPI for many years to determine how the resources allocated to local public services have changed because of inflation. The information has also been used to communicate to state legislators the needs for continued financial support to municipalities. Keeping the MPI relevant means that periodically the expenditure weights describing the types of goods and services purchased and their relative importance must be updated and that process is described in this report. There are several main findings from this project. First, the types of goods and services purchased by municipalities changed in the past 10 years but by relatively small amounts based on an analysis of expenditure summaries. As one might expect, the changes had a larger impact at the department level than on the city index. Second, based on MPI comparisons, municipalities adjusted both revenues and expenditures during the since the recession. In terms of revenues, changes in state-shared revenues coincided with increased use of property taxes which went from 18.7% of municipal revenues collected to 22.5%. Those trends are likely to continue given the current status of state finances. Third, there was some shift in expenditure patterns for services. The main change was a reduction in capital expenditures and an increase in current operating expenditures. One likely explanation is that local officials and managers, facing tight budgets, delayed replacing equipment or otherwise postponed capital projects so they could maintain current services at the appropriate level. If this is the explanation, it could suggest that in the future capital expenditures will increase if, and when, finances improve. That remains to be seen. The updated MPI can assist in local budgetary decisions especially in comparing changes in resource over time. It is relatively unique because it reflects changes in municipal purchases in Illinois. Its continued use can also communicate the need for revenues from the state and federal governments and can help understand the outcomes from not receiving these revenues. References Boskin M. 2005. Causes and Consequences of Bias in the Consumer Price Index as a Measure of the Cost of Living. Atlantic Economic Journal 33:1-13. 15

Nakamura et al. 2014. Sourcing Substitution and Related Price Index Biases. Working Papers Research Department Federal Reserve Bank of Philadelphia 1:1-56. Norman N. 2008. Producer Price Indexes: Properties, Problems and Potential Applications. The Australian Economic Review 41(4): 441-449 Park M. & Ryu D. 2011. Issues with a Chained-Type Price Index: An Analysis with the Producer Price Index. Journal of Economic Development 36(3): 47-78. Turner Construction Company (2016). Turner Building Cost Index, 2016 Fourth Quarter Forecast. Retrieved from: http://www.turnerconstruction.com/cost-index U.S. Department of Labor, Bureau of Labor Statistics (2017). Consumer Price Index. Retrieved from: https://www.bls.gov/cpi/. U.S. Department of Labor, Bureau of Labor Statistics (2017). Producer Price Index. Retrieved from: https://www.bls.gov/ppi/. U.S. Department of Labor, Bureau of Labor Statistics (2017). National Compensation Survey. Retrieved from: https://www.bls.gov/eci/. Walzer N. 1970. A Price Index for Municipal Purchases. National Tax Journal 23(4): 441-447. This price index was featured by the International City Management Association in Municipal Innovations 9 Report, May 1976. Walzer N. & Davis A. 2016. Municipal Price Index, 2015. Illinois Municipal League (April issue) download at: https://www.iml.org/page.cfm?key=7906&parent=2110. 16

Appendix. Municipalities Included in Compensation Survey The following cities were asked to provide information in the compensation survey. Asterisks indicate municipalities that were added to the survey for the 2016 MPI update. Addison Aurora Algonquin* Barrington* Beecher* Bellwood Bolingbrook Canton Carpentersville Champaign Chicago Ridge Collinsville Columbia* Danville DeKalb Decatur Des Plaines Downers Grove East Moline East Peoria Edwardsville Elgin Elk Grove Elmwood Park Evanston Galesburg* Geneva Glenview Herrin Highland Park Joliet LaSalle* Lisle* Lombard Loves Park Macomb Moline Montgomery Morris Naperville New Lenox Niles Normal Northbrook O'Fallon Oak Lawn Palatine Pekin Peoria Peru* Pontiac Quincy River Forest* Rockford* Rock Island Schaumburg Skokie South Elgin South Holland Springfield Sterling Swansea Taylorville Villa Park Wadsworth* Waukegan Wheaton Wilmette Worth 17