A Comparative Analysis of Ontario s Recycling Programs

Size: px
Start display at page:

Download "A Comparative Analysis of Ontario s Recycling Programs"

Transcription

1 CONTACT A Comparative Analysis of Ontario s Recycling Programs Bruce G. Wilson, Dept. of Civil Engineering, University of New Brunswick Dr. Bruce G. Wilson, P.Eng. Department of Civil Engineering, University of New Brunswick P.O. Box 44, Fredericton, New Brunswick, Canada, E3B 5A3 Phone: wilsonbg@unb.ca EXECUTIVE SUMMARY The first Blue Box curbside recycling program was established in the Province of Ontario, Canada in the early 198 s. Today there are approximately 8 municipal recycling programs in North America. Despite almost 25 years of experience, there is still very little definitive information available on the relative efficiency of these waste management programs. The purpose of this paper is to use statistical methods to compare the performance of recycling programs in Ontario using publicly available data on tonnages, costs, and program details. The Waste Diversion Organization of Ontario (WDO) has been collecting information on approximately 19 recycling programs in Canada s most populous province since 22. The data are collected in a standard format and are published annually on the WDO website. The data cover 144 curbside programs and 46 depot collection programs servicing between 14 and 98, households. The programs collect a range of different materials using different collection schemes at different collection frequencies. The volume of data and the differences among programs allow for statistically reliable comparisons of performance to be made. For example, the depot programs were found to collect 38% less material by weight than curbside programs. Most of this difference was due to lower tonnages of waste paper at depots. Depot collection was found to cost 51% more than curbside collection on a cost per tonne basis, however, the net costs per household for depot and curbside programs were not significantly different. This study found that the number of households served was a statistically reliable predictor of the net program cost. Weekly collection programs were found to collect 15% more material by weight than bi-weekly programs; however, there was no difference in the net cost per tonne between weekly and bi-weekly programs. Municipalities with a pay-as-you-throw program for refuse had higher average recovery of recyclables by weight, but there was no significant difference in cost between these programs and other programs. Imposing a limit on the number of bags of refuse collected at the curb was found to have no significant impact on the recovery of recyclables or net cost of the recycling program. Correlation analysis indicated that some materials are correlated with higher or lower overall program costs. For example, the weight of aluminium collected correlated well with lower net cost programs, while programs which collected polystyrene were associated with higher net program costs. Statistically significant correlations between tonnage collected and net program costs were found for a number of different materials.

2 Although the WDO provides a very good data source for comparisons among programs, the value of the data could be enhanced by collecting more detailed information on the collection programs, such as the size of the collection fleet and the total length of the collection routes in the service area. INTRODUCTION The first city-wide Blue Box recycling program in North America was established in Kitchener, Ontario, Canada in Since that time, thousands of curbside recycling programs have been established across North America and throughout that time there has been an ongoing discussion about how to pay the cost of operating these programs. Although there have been studies of the cost of collecting municipal solid waste in Canada (e.g. McDavid 2), there is very little published information on the cost and efficiency of curbside blue box collection systems. What little information there is tends to be anecdotal. Despite this lack of information, a complex system for funding recycling programs has developed in Ontario in recent years. Under the province s Waste Diversion Act (22), companies that introduce packaging and printed material into Ontario s consumer marketplace pay 5% of the net cost of Ontario's waste diversion programs. To support these subsidies, fees are collected and distributed by Stewardship Ontario, a private corporation. Stewardship Ontario collects fees from its member corporations based on the types of materials they introduce into the Ontario marketplace. Fees are set according to a formula that includes an allocation of the cost of collecting different recyclable materials. This cost allocation is based on time and motion studies of five Ontario recycling programs (Stewardship Ontario 24, 26b). In 26, total subsidies amounted to CAN$48,565,217 (Stewardship Ontario 26a). The subsidy paid to each municipality is calculated by Waste Diversion Ontario (WDO), a noncrown agency established by the Ontario Government, using a Municipal Funding Allocation Formula (WDO undated). The formula allocates funding to individual municipalities based on a number of factors including the weight and volume of material collected, the size and population density of the municipality, and average revenues from material sales. In support of this complex funding arrangement, the WDO has been collecting data on all municipal recycling programs in Ontario annually since 22. The data are collected using a standardized process, known as Generally Accepted Principles for calculating municipal solid waste system flow or GAP (CSR 22). The data are verified and posted publicly on the WDO website. The data cover 144 curbside programs and 46 depot collection programs servicing between 14 and 98, households. Despite the existence of this extensive database, very little comparative analysis of different recycling programs in Ontario has been undertaken. For example, even though the WDO has detailed information on the actual operating experience of 19 recycling programs, the costs associated with handling different materials in the blue box system are estimated based on limited information from time and motion studies from a very small subset of these programs. The hypothesis of this paper is that the WDO database contains sufficient information about Ontario s recycling programs to allow for meaningful information about these programs to be extracted using statistical techniques. If this is true, it may help establish reliable benchmarks for the efficiency of recycling collection programs and to eliminate or reduce the need to conduct labour intensive time and motion studies of sample programs to set industry fees.

3 METHODOLOGY The primary data for this research were collected by the Waste Diversion Organization through their annual Data Call process. The WDO obtains program data and tonnage reports from all Ontario municipal recycling programs. The data are collected in a standard format and are subjected to a variety of quality assurance tests (CSR 22, WDO 25). After compilation and verification, the data are made available on the WDO website. Although the data are publicly available, specific permission to use the data for research purposes was obtained from the WDO (Lance 26). This paper uses data from 24, collected by the WDO in 25. The data were downloaded in Excel format and analyzed using MINITAB 14 statistical software. Descriptive statistics, such as means, variances, and frequencies were determined for most of the parameters measured. In addition, a range of inferential statistical tests (such as t-tests, correlation coefficients, and ANOVA) were performed. In particular, the data were examined to see if they could answer the following questions: 1. Is there a difference between the efficiency and cost of curbside collection programs and depot collection ( bring ) programs? 2. Are there statistically reliable predictors of the amount of material generated by a recycling program and the cost of that program? 3. Do program variables such as the frequency of collection, the existence of a pay as you throw program, or a bag limit on household garbage have a measurable effect on the recycling collection program? and 4. Can the data be used to estimate the cost associated with collecting specific materials. RESULTS Comparison of Curbside and Depot Programs The 24 WDO database contained information on 144 curbside collection programs and 46 depot programs. Two of the curbside programs were excluded from further analysis, one because it reported a very high cost per tonne of material collected and one because it reported a very high weight of recyclable material per household served. The remaining 142 curbside programs varied greatly in size, ranging from a low of 186 households served to a high of households served. The mean program size was households and the median program size was 4178 households. Thus, the curbside data consist of a large number of relatively small programs, with a few programs serving very large cities. The size of the depot programs ranged from 14 to households, with a mean size of 2299 households and a median size of 1333 households. The data were examined using a series of two sided t-tests to see if there was a significant difference in the mean weight of recyclable material recovered by curbside and depot programs. The results are presented in Table 1. Table 1 shows that, on average, depot programs collect 58 kg/hhld (38%) less material by weight than curbside programs. These results are statistically significant at 95% confidence. However, the table also shows that approximately 1/2 of this difference is due to lower recovery of printed paper in depot programs. Depots collected 29.5 kg/hhld less printed paper than curbside programs and also collected significantly less mixed paper and paper packaging. This result is not surprising since most depot programs in Ontario serve rural areas with lower newspaper readership and residents of rural areas are more likely to burn paper products in woodstoves. The data show that depots also collect significantly less glass, PET, and HDPE, but the differences between depot and curbside collection of steel and aluminium are not statistically significant (at 95% confidence).

4 Table 1: Mean Recovery of Recyclables in Curbside and Depot Programs Mean Recovery (kg/household served) Statistical Depot Curbside Difference Significance (at Material Programs Programs (Curb-Depot) 95% Confidence) Printed Paper Yes (p=.) Residential Mixed Paper Yes (p=.) Paper Based Packaging Yes (p=.14) Flint Glass Yes (p=.) Coloured Glass Yes (p=.) PET & HDPE Yes (p=.2) Steel No (p=.396) Aluminium No (p=.412) All Other Materials No (p=.142) TOTAL Yes (p=.) Table 2 compares the net cost of depot programs and curbside programs in Canadian dollars (1 CAN$.7 ). The data show that depot collection costs CAN$131.6 per tonne (51%) more than curbside collection (p=.). This could be due to a number of factors such as lower recovery from depot programs, lower revenues for materials from depots, or higher fixed costs. Again this result is not surprising since it is generally acknowledged that depot collection is costly on an average cost per tonne basis. Table 2 does, however, show an interesting result for the average cost per household served. The net cost per household for depot collection is slightly lower (CAN$24.89 per hhld vs CAN$3.64 per hhld for curbside collection) but the difference of CAN$5.75 per household is not statistically significant at 95% confidence (p=.73). Many depot programs have been justified on the basis that, while they may cost more on a per tonne basis, they cost less than curbside collection on a per household basis. This analysis suggests that these costs savings may not be significant. Table 2: Net Program Costs for Curbside and Depot Collection (excluding 4 very large curbside programs) Parameter Program Type Sample Size Mean Std. Dev. Net Cost per Depot 46 $336.1 /tonne $339.2 /tonne Tonne Curbside 138 $222.4 /tonne $131.6 /tonne Recovered Difference $113.7 /tonne Net Cost per Depot 46 $24.89 /hhld $19.43 /hhld Household Curbside 138 $3.64 /hhld $15.83 /hhld Served Difference $ 5.75 /hhld Little more meaningful information could be gleaned from the depot collection programs, so the remainder of this paper focuses on the performance of curbside collection programs. Analysis of Curbside Collection Programs Analysis of the curbside collection programs began with a comparison of the recovery of recyclable materials by weight. The mean recovery from all programs was ± 55.7 kgs/hhld and the median recovery was kg/hhld. Recovery of materials ranged from to kg/hhld. This indicates that there is a large range of potential recovery for different programs, with the highest recovery almost three times that of the program with the lowest recovery per household. Since both the WDO and Stewardship Ontario are interested in supporting efficient collection programs, it would be useful to have a method for identifying efficient or inefficient programs. One

5 potential approach is illustrated in Figure 1, which presents a regression analysis between total households served by a program and the tonnage of recyclable material collected by that program. The figure shows that most programs are actually in a relatively narrow band of approximately the same efficiency, described by the regression equation: Total Tonnes = * Total Households (1) The correlation between households served and tonnage collected is very strong (R 2 =96.6%), meaning that the number of households in a program is an excellent predictor of the tonnage of recyclables recovered. Furthermore, Figure 1 shows that although some programs are marginally better or worse than the average, only a few programs can be identified as significantly better or worse than the average (as indicated by falling outside the 95% confidence interval). For example, Figure 1 shows that one program clearly outperforms the rest the Region of Peel collected 87,921 tonnes from 34, households for an average recovery of 259 kgs/hhld. Similarly, the City of Ottawa program collected slightly more tonnage than average, while the City of Toronto program performed below average (at kg/hhld served). Note that because of the size of the City of Toronto program, there was a concern that the regression might be affected by this one large program. When the analysis was repeated excluding the Toronto data, the results were unchanged. Similarly, repeating the analysis while excluding the 4 largest curbside programs again resulted in no significant change. 2 Total Tonnes = Households Served by Curbside Regression 95% CI 15 Toronto S R-Sq 96.6% R-Sq(adj) 96.6% Total Tonnes 1 5 Peel Ottawa Households Served by Curbside 1 Figure 1: Tonnage Recovered vs. Number of Households Served (142 Curbside Programs) The analysis presented in Figure 1 suggests that the Region of Peel program is outperforming other programs, such as the City of Toronto. However, Figure 2 presents a different view of the data that suggests the opposite may be true. Figure 2 plots the net cost of the curbside program vs. the number of households served. In this figure it can be clearly seen that the extra recovery in the Region of Peel comes at a cost. The net cost of the Peel program is significantly outside the range of normal costs and is significantly higher than the cost of other curbside programs. Conversely, the cost of the City of Toronto program is below the 95% confidence interval, suggesting that it is operating below the normal range in terms of cost per household served.

6 Figure 2 also shows that some programs appear to be significantly less expensive than other similar programs. For example, Figure 2 clearly shows that the collection program in Essex-Windsor has a net cost well below the 95% confidence interval, indicating that this program is very cost-effective. Figure 2 also presents some very useful information about the relationship between the cost of service and the number of households served by a program. The Figure shows that there is a fixed cost of approximately CAN$155, to establish a curbside collection program and that the marginal cost of providing service to an additional household is approximately CAN$ It is important to note that both the fixed and the marginal cost are statistically significant. The standard error of the marginal cost is CAN$.53, which means that we can be 95% certain that the marginal cost of curbside collection in Ontario is between CAN$18.4 and CAN$2.5 per household. Net Cost = Households Served by Curbside 2 15 Toronto Regression 95% CI S R-Sq 9.6% R-Sq(adj) 9.5% Net Cost 1 Peel 5 Essex-Windsor Households Served by Curbside 1 Figure 2 : Net Program Cost vs. Number of Households Served The results in Figure 2 suggest that the WDO formula for providing program subsidies may be overly complex. The figure suggests that simply providing each program with CAN$77,5 to cover fixed costs and an additional CAN$9.73 per household served would be an effective way of subsidizing 5% of the net cost of curbside collection programs. A similar analysis of net program cost vs. tonnage of recyclable material collected in presented in Figure 3. This figure shows that there is an excellent correlation between net cost and total tonnage (R 2 = 96.%). More importantly, Figure 3 shows less deviation from the average than was evident in either of the previous figures. For example, the Region of Peel program collected significantly more material, but also had a significantly higher net cost. The effects cancel each other with the result that Peel appears to be only slightly higher cost than one might expect. Similarly, Figure 3 shows that the City of Toronto has lower than average recovery, but it also has lower than average costs, putting it within the 95% confidence interval on a cost per tonne basis. Once again, very cost efficient programs, such as Essex-Windsor, are clearly evident.

7 Figure 3 also shows that there is a fixed cost to establishing a curbside collection program (of approximately CAN$19, and a marginal cost of CAN$113.7 per tonne of material collected. Again, both the fixed and marginal costs are statistically significant. The WDO currently subsidizes programs based on average costs, and then corrects the value for variables such as program size. The current analysis suggests that an alternative subsidy program for Ontario curbside recycling programs would be to pay them a fixed subsidy of CAN$54, and a subsidy of CAN$56.85 per tonne of material collected Net Cost = Total Tonnes Toronto Regression 95% CI S R-Sq 96.% R-Sq(adj) 96.% Net Cost 1 Peel 5 Essex-Windsor Total Tonnes Figure 3: Net Program Cost vs. Total Tonnage Recovered Estimating the Effect of Program Variables There are a number of differences in the operations of the various curbside programs in Ontario. For example, some collect recyclables from each household on a weekly basis, while others collect on a biweekly basis. Some programs have attempted to increase the recovery of recyclables by instituting a pay-per-use or pay as you throw charge on household garbage collection. Others have set limits on the number of bags or containers of garbage that a homeowner can set out for collection, again in an attempt to divert more material into the recycling collection stream. The purpose of this section is to examine the effectiveness of these three program differences on the recovery of materials and the cost of the program. Frequency of Collection The WDO reports that 72 curbside programs collect recyclables on a weekly basis, while 64 programs collect biweekly. Six programs that collect with a different frequency were excluded from the analysis due to a small sample size. Table 3 shows that the weekly programs collected 21.7 kg/hhld (15%) more material than the bi-weekly programs. The results were significant at 95% confidence (p=.24). Despite this difference in recovery, there was no significant difference in the net cost per tonne. The net cost per household was CAN$4.3 (15%) more for weekly collection, but this difference was not statistically significant at 95% confidence. The results in Table 3 suggest that more frequent collection does result in higher recovery of recyclable materials at approximately the same cost per tonne as biweekly collection. The cost per household is slightly higher for the higher level of service.

8 Table 3: Effect of Collection Frequency on Recovery and Program Cost Collection Frequency Number of Programs Recovery (kg/hhld) Net Cost/tonne (CAN$/tonne) Net Cost/hhld (CAN$/hhld) Weekly (1) Biweekly (2) Difference (1)-(2) p-value Pay as you throw The WDO reports that 83 curbside programs are associated with some form of pay as you throw program for garbage collection, while 59 programs do not charge a fee per bag or container of garbage. Table 4 shows that the pay as you throw programs collected 25.7 kg/hhld (18%) more material than programs without a direct fee. The results were significant at 95% confidence (p=.5). Pay as you throw programs had a lower net cost per tonne of recyclables collected and slightly higher cost per household served, but in both cases, the difference was not significant. The results in Table 4 suggest that the existence of a pay as you throw fee on garbage collection results in a significant increase in the recovery of recyclable material, without significant increases in cost. This finding agrees with the results of previous research (Dijkgraaf and Gradus, 24). Table 4: Effect of Pay as you throw on Recovery and Program Cost Pay as you throw Program? Number of Programs Recovery (kg/hhld) Net Cost/tonne (CAN$/tonne) Net Cost/hhld (CAN$/hhld) Yes (1) No (2) Difference (1)-(2) p-value Bag Limit The WDO reports that 72 curbside programs are associated with garbage collection programs that limit the number of bags or containers that any one household can set out for collection. Another 7 programs had no limits on the number of bags or containers of garbage. Table 5 shows that the existence of a bag limit has no measurable effect on the curbside recycling program. There was no significant difference in the recovery of material, net cost per tonne, or net cost per household served. Table 5: Effect of Bag Limits on Recovery and Program Cost Bag Limit? Number of Programs Recovery (kg/hhld) Net Cost/tonne (CAN$/tonne) Net Cost/hhld (CAN$/hhld) Yes (1) No (2) Difference (1)-(2) p-value Effect of Different Materials on Program Cost The WDO data include the weight of different materials collected by different programs. Since the gross and net costs of those programs are also reported, it is possible to examine the effect of collecting more or less of any particular material on program cost. Table 6 reports the results of a stepwise regression analysis of gross and net costs per tonne with different materials. In this table, negative numbers mean that the material is correlated with a reduction in program cost, while positive numbers indicate tat the material is associated with higher program costs. Those

9 coefficients that are significant at 95% confidence are shaded. For example, Printed paper does not appear to have a significant effect on gross program costs, but printed paper does appear to reduce the net cost of a curbside collection program by about CAN$95.22 per tonne of printed paper collected. It is important to note that these cost coefficients are averages across a wide range of programs. They also include all program costs (including collection, processing, and administration). The net costs include the revenue from the sale of the material. Some caution should be exercised in interpreting these results. For example, while the implication that collecting more aluminium will reduce program costs because of high revenues, the suggestion that including more plastic tubs and lids will do the same does not agree with the field experience of many program operators. The important result here is that statistically significant difference between different materials can be measured in the existing data and that additional data and analysis could give even better results. This suggests that the current practice of estimating the cost of handling each material through a relatively small time and motion study could be replaced with statistical techniques using data from all programs. Table 6: Effect of Material Type on Gross and Net Program Cost Material Effect on Gross Cost p-value Effect on Net Cost p-value (CAN$/tonne of (CAN$/tonne of material collected) material collected) Printed Paper Paper-Based Packaging Residential Mixed Paper Polycoat Aluminum Steel Mixed Metal Flint Glass Containers Coloured GlassContainers Other Eligible Glass PET HDPE Plastic Film Tubs & Lids Polystyrene Mixed Plastics Commingled Discussion The WDO data provide considerable insight into the performance of Ontario s recycling programs. The data can be used to compare the efficiency of different programs and could be used to establish the subsidies for programs and to estimate the relative cost of including different materials in collection programs. The use of the full data set for all collection programs might be an improvement over using small samples from a few programs as a basis for estimating costs, subsidies, and material fees. Limitations The analysis presented in this paper was for the year 24. The analysis should be repeated for other years. This analysis could yield different results.

10 Conclusions and Recommendations Given the results presented above, the WDO should reconsider the use of average costs (per tonne or per household) when comparing different programs. The data clearly show that there is a fixed cost associated with establishing a program and a marginal cost associated with collecting additional material. These costs are statistically significant with a high level of confidence. Average cost per tonne is highly variable primarily because of variability in the size of programs. The WDO should also consider expanding the range of the data that they collect on each program. For example, they could consider collecting data on the number of vehicles used in each collection program, the type of collection vehicles used (manually loaded or not), the distance travelled by those vehicles in the year, the number of personnel employed or the total number of hours worked, the spatial extent of the collection area in hectares, and the tonnage of garbage collected in the municipality. Collection of this additional data would make a valuable database even more useful for comparing the performance of different recycling programs. Acknowledgements Although the WDO data are publicly available, the permission of the WDO to use the data is gratefully acknowledged. References CSR (22). GAP: A Protocol to Measure Municipal Solid Waste: GAP 2 Manual: Municipal Waste Flow. First Release. November 22. Corporations in Support of Recycling, Toronto. Dijkgraaf E. and Gradus R.H.J.M. (24). Cost savings in unit-based pricing of household waste: The case of The Netherlands. Resource and Energy Economics, 26, Lance, R. (26). Personal Communication with Ron Lance, Data Manager, Waste Diversion Ontario. May 31, 26. McDavid, J.C. (2). Alternative Service Delivery in Canadian Local Governments: The Costs of Producing Solid Waste Management Services, Canadian J. of Regional Science 23:1, Stewardship Ontario (24). Blue Box Materials Cost Allocation Study, Final Report March Report Cost Allocation Study.pdf, June 2, 27. Stewardship Ontario (26a). Annual Report June 2, 27. Stewardship Ontario (26b). Methodology for Calculating Blue Box Steward Fees. June 2, 27. Waste Diversion Act (22). June 2, 27. Waste Diversion Organization (undated). Municipal Funding Allocation Model. MFAM description for 26 posting(1).pdf, June 2, 27. Waste Diversion Organization (25). Waste Diversion Ontario 25 Annual Report, June 2, 27.