Printer Friendly

Recycling in multifamily dwellings: does convenience matter?

Factors that decrease the the cost of recycling have significant positive correlations with recycling rates in multifamily dwellings (MFDs). Waste-management experts have previously used anecdotes to infer a link between convenience and waste-diversion rates in MFD recycling programs. This article confirms and quantifies that link by applying probit and double-censored to bit analysis to survey data from 214 households in Urbana, Illinois. We find a strong connection between recycling rates and the perceived presence of adequate interior space for processing recyclables, and distance to recycling bins affects container-recycling intensity. The results have implications for cost-effective design of MFD recycling programs. (JEL DI, Q2)

I. INTRODUCTION

Statistics from the U.S. Environmental Protection Agency (EPA) show that from 1960 to 1997, U.S. recycling rates rose from 6.4% to 28%, and 9,000 curbside recycling programs were established (EPA 1998). Such programs have traditionally been geared toward single-family dwellings (SFDs); conventional doctrine has been that recycling activity is likely to be limited in multifamily dwellings (MFDs). However, as residential recycling rates for SFDs plateau in the United States, solid-waste managers and policy makers are becoming increasingly interested in MFD recycling programs. By now, at least half of all MFDs may have access to a curbside recycling program (EPA 2001). Stevens (1999) and the EPA (1999, 2001) show that waste-diversion rates (the fraction of a household's waste that is recycled instead of discarded as refuse) tend to be slightly lower in MFDs than in SFDs, but some MFD recycling programs have yielded diversion rates as high as 60%.

What factors influence the recycling rates chosen by residents of MFDs? In a detailed nationwide study, the EPA (2001) found that "successful" (20% or higher diversion rate) MFD recycling programs tended to have large and numerous containers provided by the recycling service. Waste-management experts emphasize the importance of convenience in yielding high waste-diversion rates in MFD recycling programs, with some "record-setting" programs using strategies like doorstep pickup of recyclables in high-rise apartment buildings (EPA 1999).

Such anecdotes are intriguing. However, although much work has studied SFD recycling behavior, little has been done to explore MFD recycling in a rigorous statistical fashion. A number of economists have studied the sensitivity of recycling behavior to pricing policies that alter the costs and benefits of recycling relative to disposal. (1) Another body of work has investigated the influence of nonpecuniary policies and variables on recycling in SFDs. Work by Vining and Ebreo (1990) and Ebreo and Vining (2000) supports the notion that convenience may increase the number of people that identify themselves as recyclers. However, their data are strictly qualitative; the earlier study dates from a time when only community drop-off recycling was available, and the later study is focused on public attitudes toward and expectations regarding changes in recycling policy. Jakus et al. (1996, 1997) use economic models and rigorous empirical analysis to study the influence on actual recycling behavior of recycling promotional campaigns and policies to alter the time cost of recycling. They find evidence that recycling participation is not a function of home ownership or renting status per se; however, these articles do not delve into the details of MFD recycling because the data were collected in the context of rural drop-off recycling.

A few studies have turned their attention to MFD recycling. Stevens (1999) finds that programs with high aggregate diversion rates are likely to have a relatively large ratio of collection bins to households. McQuaid and Murdoch (1996) conduct a qualitative analysis of MFD recycling in Edinburgh that suggests that housing characteristics, such as floor level, presence of an elevator, and household size, may influence recycling rates. Westergard (1996) finds in one community that willingness-to-pay for a MFD recycling program is positively correlated with the amount of space households have to store recyclables and negatively influenced by the time constraints that households face.

Taken together, the current literature suggests that households in MFDs may recycle more when the transaction costs associated with storing and appropriately discarding recyclables are reduced. This article subjects that hypothesis to a rigorous empirical test using household-level data on recycling rates, recycling-bin distances, and perceived storage space in MFDs SFDs in a Midwestern community.

II. BACKGROUND ON RECYCLING AND TRASH COLLECTION IN URBANA (2)

The recycling program found in Urbana, Illinois, is not atypical of programs found in other communities its size. Urbana is a city of about 35,000 people, the demography of which is highly influenced by the presence of a large university. In 1986, the City of Urbana began a city-wide program, named U-Cycle, to provide curbside collection of recyclable materials from all SFDs within the city limits. In August 2000, U-Cycle expanded to include an MFD recycling program as well. An MFD is defined as a building having seven or more dwelling units; this definition encompasses apartments, rooming houses, fraternities, sororities, dormitories, and condominiums. Remaining dwellings of fewer than seven units are defined as SFDs.

Though all landowners in Urbana pay for U-Cycle through property taxes and fixed fees, recycling participation is voluntary. The SFD recycling program provides weekly collection of two categories of recyclables. The first category is paper fibers, including newspaper, cardboard, magazines, junk mail, and paperboard. The second category is containers, including steel, tin, aerosol cans, glass bottles, and plastic with PETE 1 or HDPE 2 recycling labels. Each household is provided two small, 35-gallon bins to set out on the curb for collection. Participants are required to sort their recyclables by category, with one category of materials going into each of the bins provided. Residents must carry the bins out on the curb for collection day, but the bins cannot be stored there.

MFDs have access to communal 95-gallon recycling collection bins, usually located in the nearest open space by the building. Residents of such dwellings may deposit recyclables in those bins at any time, though the receptacles are emptied by the city only once a week. MFD households must sort their recyclables into roughly the same two categories as the SFD households; (3) each receptacle is clearly labeled with a list of materials that may be placed into it.

To understand recycling behavior, we need to understand something about the institutions that govern household trash collection as well. Residents of MFDs do not pay for waste disposal directly. Instead, they deposit their trash into ground-floor communal dumpsters provided by the property owner. Those dumpsters are usually close to the recycling bins, so carrying a bag of trash to the dumpster requires approximately as much effort as carrying a bag of recycling to the bins.

Residents of SFDs are not provided with uniform waste disposal services arranged by a central city authority. Rather, they are free to choose from a number of different licensed waste haulers and may choose contracts with different fee structures and amenities. It is possible to arrange at little cost to have one's trash picked up without having to haul it to the curb--so-called backdoor pickup. Thus, SFD inhabitants can arrange for trash disposal to be much more convenient than recycling. All contracts have fixed monthly fees; a few (but not all) of those fees are $1-2 more costly per can of trash the hauler is to collect. For example, if having one can collected each week costs $12, having two cans collected each week might cost $14. Recycling will only influence trash disposal costs for a SFD homeowner if the quantity of waste diverted is enough to reduce the number of big cans the homeowner needs to have collected each week--an unlikely event.

III. SURVEY AND DATA

Because MFD households share communal waste and recycling bins into which they may deposit materials at any time, objective measurement of household-level waste-disposal activity is nearly impossible. Thus, to gather data for an analysis of the impact of convenience on recycling behavior, we conducted a survey of 155 SFDs and 496 units in MFDs in Urbana. A written survey instrument was delivered by hand to the front door of each surveyed household, with a self-addressed envelope included to facilitate response; that questionnaire is replicated in the appendix. The written instrument elicits data on four different recycling rates: the percent of total waste the household recycles, the percent of newspaper waste they recycle, the percent of recyclable containers they actually recycle, and the percent of recyclable nonnewspaper paper fiber they actually recycle. It asks whether respondents consider recycling to be convenient, whether they have adequate space for storing recyclables, and whether they recycle when they are not at home (e.g., throwing soda cans into public recycling receptacles). The survey also includes questions about household size and gender composition and the age, education, and work/student status of the respondent. (4) Many more questions were asked in the survey than are included in the present analysis; in many cases (e.g., presence of disabled people, presence of children) there were too few affirmative answers to yield a variable with adequate variation.

When each survey was distributed, information was recorded by the surveyor about the physical situation of the dwelling. This includes the distance from each household's door to where they deposit their recyclables, (5) on how many floors above ground level the front door is located, and whether the building (in the case of MFDs) has a functioning elevator. Distance to recycling location was measured using a pedometer and then converted into hundreds of feet. (6)

Households were sampled using a stratified random technique. We divided the city of Urbana into four areas based on the boundaries of voting districts, and surveyed the same numbers of houses and apartment complexes in each quadrant. We aimed to concentrate the surveys of apartments in a relatively small number of complexes, so that fixed effects could be used to control for unobserved heterogeneity at the building level. (7) There are two sources of deviation from randomness in the selection of MFDs. First, there is a bias toward taller buildings. Urbana is dominated by buildings that have four or fewer floors; (8) thus, to maximize variation in the floor on which apartments in our sample were located, we only surveyed MFD units in three- or four-story buildings. Second, the third of three waves of survey distribution was targeted toward neighborhoods that were likely not to be dominated by university faculty and graduate students in an effort to expand the variation in educational attainment in our sample. (9)

The survey response rate was 33%, yielding 217 observations; this is consistent with the 29% response rate to Ebreo and Vining's (2000) hand-delivered survey. Three of the households reported no information about recycling, bringing the effective sample size to 214. Of those observations, 71 are responses from SFD households. The other 143 observations are responses from the occupants of apartments, and those apartments are distributed among 13 different apartment complexes. Only 23 of the MFD households get a daily paper; thus, this article does not analyze newspaper recycling rates. Table 1 provides summary statistics for the variables we use in this article, with the data broken down into subsamples by dwelling type. Our sample seems reasonably representative of Urbana in terms of age, but our responding households seem to have slightly fewer occupants than reported by the Census, and our sample contains a disproportionately large fraction of people with bachelor's degrees or higher.

Reported total diversion rates are strikingly high--averaging 0.31 for apartments and 0.47 for houses in our sample and variable, with standard deviations typically larger than half the size of the means. The city itself does not report volume-based diversion rates that are directly comparable to the measure elicited in our survey (our questions encouraged respondents to think about waste disposal in terms of numbers of bags). However, U-Cycle officials claim that average mass-based total diversion rates are around 11.7% (Rushforth 2003). If mass- and volume-based measures are comparable, then the reported rates in our sample are high. That could be due in part to recall error--respondents thinking themselves to be more prolific recyclers than they really are. It may also be the product of sample-selection bias, if well-educated recycling enthusiasts are more likely to return the survey. Either of those two biases is likely to act in similar ways for the two subgroups in the sample, yet we do observe that apartment dwellers report lower recycling rates than do residents of SFDs.

What lies behind this pattern? It could be that inhabitants of MFDs feel less ownership of their own waste stream and thus gain less utility from reducing it (though respondents from the samples report very similar propensities to recycle when they are not at home). Apartment dwellers might also recycle less because they do not pay for their trash disposal directly, whereas occupants of SFDs must contract with a waste hauler to dispose of any waste they do not recycle (though because the marginal cost of trash hauling is essentially zero, this effect is probably small).

The differences in recycling rates might also be driven by other differences between housing types and their occupants. Table 1 shows that the lateral distance from front door to bin location (measured as total distance less the distance traveled on stairs) is more than twice as great for apartments, and apartments are, by definition, more likely to be located above the ground floor. SFD respondents are much more likely to feel that they have enough space to store recyclables. Apartment dwellers are younger, more likely to have work and study commitments that yield more than a full-time workload, and much less likely to subscribe to a newspaper, and the gender composition of apartments is much less likely to be mixed. Only a statistical analysis can sort out which of these factors drive the variation in reported recycling rates that we observe.

IV. ANALYTICAL FRAMEWORKS

The work by Jakus et al. (1996, 1997) analyzes determinants of whether households can be categorized as recycling participants; the earlier work also studies variation in the volume of materials recycled, but not as a fraction of total waste. This work captures important dimensions of household waste-disposal behavior but does not permit analysis of the extent to which independent variables might change the intensity of recycling activity in which recycling participants engage. Our data reveal much variation in recycling intensity in between all and nothing. Hence, our empirical work extends the literature by studying the influence of nonpecuniary factors on recycling rates.

The histograms of Figure 1 show the distributions of the three reported recycling rates that comprise our dependent variables: the fraction of total waste that is recycled, the fraction of recyclable nonnewspaper paper that is recycled, and the fraction of recyclable containers that are recycled. The distributions of all three variables are censored at zero and have sizable point masses there. The distributions of paper and container recycling rates also have large point masses at one. Such dependent variables can not be appropriately handled by performing an ordinary least squares analysis of reported recycling rates. Thus, we employ two nonlinear frameworks for our analysis of household recycling behavior.

[FIGURE 1 OMITTED]

For both analyses, it is important to remember that in Urbana, as in many other communities in the United States, the private pecuniary marginal cost of trash disposal is effectively zero. Voluntary recycling programs like U-Cycle are structured such that the private pecuniary marginal cost of recycling is also equal to zero. Variation in recycling rates must therefore stem from variation in the nonmonetary costs of recycling and in the nonpecuniary benefits households gain from engaging in this activity. In particular, theoretical work such as that found in Morris and Holthausen (1994) and Jakus et al. (1996) lead us to expect a household's propensity to recycle to decrease with the time constraints a household faces, and with the time cost and added effort associated with choosing to recycle something rather than disposing of it with the rest of the trash. We also expect recycling rates to increase with the psychic benefit the household obtains from recycling instead of discarding waste, which may be a function of ideological beliefs, age, education, and other demographic characteristics. We include independent variables in the analyses to capture as many of these factors as possible.

Model 1

We seek to use an econometric estimator that exploits the information contained by responses between zero (no recycling) and one (recycling 100%), while accommodating the fact that there are substantial point masses at zero and one in the distributions of the dependent variables. Because of those point masses, logit or probit regression for grouped data is not appropriate. (10) Instead, we use a double-censored tobit method shown in Greene (2000, p. 906), because that method is particularly well suited for cases where the limit observations are numerous and qualitatively different from the nonlimit observations.

A double-censored tobit estimator can be inspired by the following conceptual framework. Suppose household h has a propensity to recycle, [Y.sup.*.sub.h], that increases with the utility the household members get from having high recycling rates and decreases with the costs they must incur to recycle a given fraction of their waste. Because we expect those costs and benefits to be functions of characteristics of the residents and of the dwelling itself, the propensity to recycle will also be a function of those variables, [X.sub.h], according to:

(1) [Y.sup.*.sub.h] = [[beta].sub.1] + [[epsilon].sub.h],

where [[beta].sub.1] is a vector of parameters to be estimated by model 1 and [[epsilon].sub.h], is a normally distributed error term with mean equal to zero. Given the natural constraints on the range of feasible recycling rates, we observe rate [Y.sub.h] that is related to [Y.sup.*.sub.h] as follows:

(2) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]

We estimate this model using a standard double-censored tobit analysis.

Model 2

The second analytical framework is similar to the analyses performed in Jakus et al. (1997) and Vining and Ebreo (1990). This approach quantifies the effects that convenience and other factors have on whether a household recycles at all. To that end, we generate discrete dependent variables equal to one if the relevant recycling rate is greater than zero and zero if not, discarding information on differences in recycling intensity among recycling participants. We then perform standard probit analyses of these discrete dependent variables.

V. RESULTS

The econometric results are reported in Tables 2 and 3. The analyses performed in conjunction with the double-censored tobit (model 1) were run first with 13 building-specific dummy variables. Those dummies were dropped from regressions in which they were not significant, and results from the more parsimonious specification are reported in those cases. The results will be discussed one model at a time; a discussion of the size of marginal effects follows discussion of parameter estimates.

Model 1

Table 2 presents double-censored tobit analyses of recycling rates. The table reports the coefficient values in the equation that determines the latent propensity to recycle. The first three columns are results from the full sample of houses and apartments. Factors other than housing type do appear to have significant explanatory power over recycling rates. However, at least some of the gap apparent in the sample means of recycling rates is significantly correlated with a difference between houses and apartments not captured by other explanatory variables. It may be that SFD dwellers have, on average, higher income, which has been shown in other studies to be positively correlated with recycling.

Some of the other results in these columns warrant our attention because they highlight some key differences between houses and apartments. These analyses allow for the impact of convenience to be different for houses than for apartments (both convenience variables are included alone and interacted with a dummy variable for whether the observation is a house). We find that although apartment dwellers are likely to recycle more if they have adequate storage space, the presence of adequate interior storage space is not significantly correlated with recycling activity in SFDs. (11) This is likely to reflect the fact that residents of SFDs have access to exterior storage space for their recyclables (e.g., garages, porches, back yards); such space tends not to be available for use by MFD inhabitants. Thus, interior storage space has a more potent impact on recycling activity in MFDs.

In contrast, the distance recyclers must carry their recyclables has the same pattern of influence in SFDs as in MFDs (we discuss the nature of that pattern later in this section). On one hand, we might have expected bin distance to be a more important factor in SFDs because occupants of SFDs must take recycling to the curb, whereas they can arrange for backdoor pickup of their trash. On the other hand, the statistics in Table 1 make clear that bin distances are both smaller and less variable in SFDs than in MFDs, which might mean that relatively little of the variation in SFD recycling rates can be explained by variation in bin distance. Either neither of these factors is important (for example, many SFD residents do not arrange for backdoor trash pickup) or the two effects cancel each other out in our data.

Before leaving the full-sample results, we note that recycling activity seems to be positively correlated with age and education, a finding that has been previously reported in the literature. Newspaper subscribers also have higher overall waste diversion rates, perhaps because newspaper subscribers have, on average, relatively high incomes or because having a newspaper subscription yields a large volume of easily recycled paper that spurs households to overcome the startup costs of developing a recycling routine. Those results are less apparent when we drop SFDs from the sample. This may be an artifact of the reduction in sample size. It may also be that variation in the education of MFD respondents is correlated with the MFD fixed effects.

We now turn now to the findings in the latter three columns of Table 2, where the determinants of recycling rates are estimated on a sample restricted to MFDs. Several convenience-related factors are correlated with recycling rates. Respondents who have adequate space to store and sort their recycling have higher recycling rates across the board, a pattern that is most pronounced for bulky containers. This finding is consistent with the results in Jakus et al. (1996, 1997). Because this space variable is self-reported, (12) it could confound the influences of physical interior space and attitudes toward recycling; people might be more likely to report that they do not have adequate space to recycle if they are generally disinclined to recycle. However, attitudes toward recycling are controlled for

somewhat by the variable that indicates whether respondents engage in recycling when they are not at home. The coefficient on that variable is positive and significant in at least the equation for paper recycling rates, confirming our expectation that people who are committed enough to recycle when out in public are likely to have higher household recycling rates as well. (13)

Results are mixed for the convenience variables that capture variation in effort required to take recycling to the bins. The lateral distance that recyclers must walk to put out their recycling has a significant negative effect on container recycling but not on paper recycling or overall diversion rates. We can understand this effect by remembering that recycling bins and dumpsters are located near each other. Thus, we only expect bin distance to have an effect on MFD recycling if by choosing to recycle an item of waste one commits to taking an extra trip to the waste-disposal area--a trip that would not be necessary if the item were carried out in a bag along with the rest of the trash. Containers are bulky, messy, heavy, and (in some cases) fragile and not readily tucked under one's arm for a trip to the recycling bin en route to carrying the rest of the trash out to the dumpster. Thus, it is plausible that bin distance could depress container recycling but not other forms of recycling activity.

The estimated relationships between recycling rates and vertical distance from the bins are surprising. Households on the second and third floors have recycling rates that are largely the same as those of households on the first floor (second-floor residents even have statistically higher paper recycling rates), and fourth-floor residents actually appear to recycle more than those on other floors. These results are at odds with the finding that lateral bin distance lowers container-recycling rates. This may be an artifact of the data; there are, for example, only 15 households on the fourth floor in our sample. The perverse results might disappear in a much larger sample with greater variation in floor.

Other variables that capture features of the households' residents are sporadically significant. Respondents that go to the trouble of recycling when they are not at home have higher recycling rates for other paper, perhaps because they gain more utility from waste reduction than do other respondents. Paper recycling rates are also higher for respondents who have work and study responsibilities that sum to less than full-time commitment. Such individuals may be less time-constrained and thus more likely to spend the time needed to conduct triage on junk mail and flatten paperboard boxes for recycling. Households with two adults seem to have higher recycling rates for containers; the presence of another adult may increase the household's total endowment of time available for performing such chores as recycling disposal. Single-gendered households and homes with older occupants report higher total diversion rates, and women seem to recycle a much greater fraction of their containers.

Model 2

Because all the SFD households in our sample report nonzero recycling rates for total waste and for containers and only two report that they do not recycle paper, our probit analysis of recycling participation is limited to the sample of MFD households. That analysis is further complicated by the fact that the 84% of households in the MFD sample report nonzero recycling rates. This low variation in the dummy dependent variables means that some of the independent variables must either be transformed (in the case of the measure of the numbers of adults in the household) or excluded from the analysis due to excessive correlation with the dependent variables. Thus in Table 3 we report the results of a set of somewhat limited probit analyses. The table reports the marginal effects of the covariates on the probability that a household reports nonzero recycling activity.

We find that floor and bin distance are not significantly correlated with whether a household engages in recycling. However, just as in the tobit analyses, there is a strong positive correlation here between reports of adequate interior space and reported recycling participation. MFD households with enough interior space to sort and store their recyclables are 10%, 12%, and 12% more likely to report nonzero total recycling rates, paper recycling rates, and container-recycling rates, respectively.

The results for other variables are similar to those found in the tobit analysis. Respondents with less intense work and study responsibilities are 16% more likely to recycle paper and 6% more likely to recycle at least some of their containers. Households that report that they recycle when they are not at home are 18% more likely to report some paper recycling. Having more adults in the household increases the likelihood of reporting some overall recy cling and container recycling. Years of education is positively correlated with the probability that a respondent engages in container recycling. Having a single-gendered household increases the likelihood that the household has nonzero reported container-recycling rates (though all-male households are less likely to report any paper recycling).

Size of Effects

Table 4 provides calculations based on the tobit results to illustrate the approximate size of the effects of perceived interior space and bin distance on recycling rates. Note that the effects of bin distance are significant only for container-recycling rates. All three recycling rates are inelastic with respect to these convenience variables, but the statistically significant effects are nontrivial from a policy perspective.

Because bin distance is continuous, it is easiest to evaluate its importance by looking at the elasticity estimate. The first row of Table 4 implies that if bin distances were cut by 100% (perhaps reflecting the implementation of doorstep collection) recycling rates for containers would rise by about 66% of their previous levels. We can best evaluate the importance to individual recycling rates of having adequate space, which is captured here only as a dummy variable, by considering the marginal effect of that variable. The creation of adequate space (for example, by eliminating the requirement that households sort their recyclables into two categories) can increase the diversion rate of an average household by 0.17. The marginal effect of space is similar in magnitude for paper recycling rates (0.19), but much larger for containers (0.43). This pattern is sensible, because containers are messy and bulky, and take up much more interior storage space than do most types of recyclable paper.

VI. CONCLUSION

In the community we studied, apartment dwellers reported lower recycling rates than inhabitants of SFDs. The gap does not, however, seem to be just a result of the fact that apartment dweller have less of a sense of owning their own waste stream. Rather, housing-type differences in recycling convenience and occupant demographics play important roles in forming recycling rates.

Several elements of convenience influence recycling behavior in MFDs. First, recycling rates are higher in households that report having adequate interior space available for sorting and storing recyclables--as much as 0.43 points higher for the average household. Indeed, MFD recycling is more sensitive to the dearth of interior space than is recycling in single-family homes, probably because SFD residents can exploit exterior space for recycling purposes. All else equal, waste managers might reasonably expect that any given MFD recycling program will have higher diversion rates and yield higher net benefits to participants, in places where the apartments themselves are more spacious. Although policy makers who aim to increase recycling rates cannot change the size of apartment units, policy efforts might be able to influence some of the factors that make up perceptions of adequate space. Perceptions of space adequacy may be related in part to attitudes toward recycling; education and promotion drives might motivate participants to find the space they have to be more sufficient than they thought previously. Policy makers might also be able to reduce the amount of interior space that is needed for sorting and storing recyclable waste through innovations such as providing MFD residents with a single indoor storage container that has separate compartments for recyclables and trash or reducing the number of streams into which recyclables must be sorted.

Second, although the convenience of recycling bins does not have a significant effect on whether MFD households report nonzero levels of recycling activity, the rate at which MFD households recycle containers is significantly negatively correlated with the lateral distance that participants must travel to recycling bins. The elasticity of container recycling rates with respect to distance from the bins is as high as -0.66. It seems that MFD dwellers are less able to combine trips to take out trash and recycling if the recycled materials are containers instead of paper. Thus, container recycling is sensitive to the distance involved in that extra trip to the bins, because distance increases the price a resident pays in the form of time spent and effort exerted to recycle instead of discard. It may be that all recycling rates would rise more dramatically if changes were made to reduce the distance of recycling bins relative to the distance of trash disposal facilities. (14)

This article is the first to analyze the impact on MFD recycling of changing aspects of convenience that shape the household production function for waste disposal; it should not be the last. More work is needed to quantify the extent to which variance in relative convenience of recycling and trash disposal really does increase recycling rates. It would also be useful to study MFD recycling in different types of communities. For example, Urbana has no high-rise apartment buildings; there may be a nonlinear relationship between recycling rates and the number of floors between an apartment and the recycling bins that cannot be identified with the data available to us. Finally, our study has focused just on recycling behavior and, implicitly, the benefits to apartment dwellers of recycling programs. Future research should, however, investigate whether the benefits (to households and the city) of convenience improvements that increase MFD recycling rates are worth the municipal costs of making it easier for these households to be green.

APPENDIX: SURVEY QUESTIONNAIRE

The questionnaire was formatted to fit on a single side of a sheet of paper to encourage households to respond.

1. Do you feel it is convenient for you to recycle your household waste?

(Check one) Yes -- No --

2. Do you feel you have adequate space in your household to store your recyclables?

(Check one) Yes -- No --

3. Do you recycle items such as pop cans, newspaper, or glass bottles when you are not at home?

(Check one) Yes -- No --

4. Do you get any daily newspapers at home?

(Check one) Yes -- No --

If so, how many different newspapers do you get? (For example: 1 if just the News-Gazette) --

5. Take a moment to think about how much waste your household produces in a week. What percentage of that waste do you recycle? (For example, if you have approximately 1 bag of garbage a week, and 1 bag of recycling a week, you would be recycling 50% of your waste.) -- %

6. It is often hard to recycle all materials that could be recycled.

Newspaper can be recycled. In a given week, what percentage of your newspaper waste do you recycle or set aside to be recycled? (Enter NA if you have no such waste.) -- %

Aside from newspaper, some of the other paper products that you use can be recycled, such as envelopes, scratch paper, cardboard, poster board, paper packaging, magazines and paper bags. In a given week, what percentage of the recyclable paper waste that you generate do you recycle or set aside to be recycled? --%

Some of the containers that you use can be recycled, such as glass bottles, aluminum cans, tin cans, plastic containers and steel. In a given week, what percentage of the recyclable container waste that you generate do you recycle or set aside to be recycled? -- %

7. Which description fits this residence:

(Check one) Owner-occupied -- Rental --

8. Your household is:

(Check one) Mixed gender -- All female -- All male --

9. Including yourself, how many adults (ages 18 and above) live in this household?

How many children (aged less than 18) live in this household? --

What is the age of the oldest adult? -- years

What is the age of the youngest child? -- years

10. Does any adult in the household have a physical disability that makes carrying trash and recycling bins difficult for him or her?

(Check one) Yes -- No --

If yes, how many adults meet that description?

11. Check the description(s) that best describe you (you may want to check more than one):

Student enrolled part time --

Student enrolled full time --

Employed part time (1-20 hours/week) --

Employed full time (21-40 hours/week) --

Employed more than full time (41 + hours/week) --

Other --

12. Circle the highest year of education you have completed:

8 9 10 11 12 13 14 15 16 17 18 19 20
TABLE 1
Mean Values of Variables (a)

 MFDs SFDs

Fraction total waste recycled 0.31 (0.22) 0.47 (0.21)
Fraction paper recycled 0.50 (0.39) 0.71 (0.32)
Fraction containers recycled 0.60 (0.39) 0.83 (0.29)
Floor (1 = ground) 2.3 (0.94) 1.0 (0.0)
Lateral distance to bins (100 ft) 1.6 (0.82) 0.60 (0.22)
Education of respondent (years) 17 (2.2) 16 (2.5)
 High school graduate? 0.98 0.96
 Bachelor's degree or higher? 0.70 0.64
Age of oldest person (c) 29 (11) 48 (20)
Answered age? 0.85 0.87
Number of people in household 1.6 (0.69) 1.8 (0.77)
 2 adults in household? 0.45 0.46
 > 2 adults in household? 0.08 0.16
Enough space to store recyclables? 0.51 0.83
Recycle tm ay from home? 0.70 0.67
Subscribe to a newspaper? 0.16 0.58
Less than full-time? (d) 0.12 0.33
More than full-time? (e) 0.40 0.23
All male? 0.34 0.10
All female? 0.40 0.15

 Full Sample Census (b)

Fraction total waste recycled 0.37 (0.23)
Fraction paper recycled 0.57 (0.38)
Fraction containers recycled 0.68 (038)
Floor (1 = ground) 1.9 (0.99)
Lateral distance to bins (100 ft) 1.5 (0.93)
Education of respondent (years) 17 (2.3)
 High school graduate? 0.98 0.91
 Bachelor's degree or higher? 0.68 0.54
Age of oldest person (c) median = 26 median = 25
Answered age? 0.86
Number of people in household 1.9 (0.98) 2.14
 2 adults in household? 0.45
 > 2 adults in household? 0.11
Enough space to store recyclables? 0.62
Recycle tm ay from home? 0.69
Subscribe to a newspaper? 0.30
Less than full-time? (d) 0.19
More than full-time? (e) 0.34
All male? 0.26
All female? 0.33

(a) N = 143 for MFDs: N = 71 for SFDs: SD in parentheses.

(b) U.S. Census Bureau (2000).

(c) In regressions, "Age of oldest person" equals that age if
not missing, and zero otherwise. Here, the means of age are
calculated only over nonmissing replies. The reported medians
are calculated across adults and children.

(d) "Less than full-time?" is a dummy for study and work
duties summing to less than full-time.

(e) "More than full-time? is a dummy for study and work
duties summing to more than full-time.

TABLE 2
Results of Tobit Regressions on Reported Recycling Rates

 All Observations

Sample Dependent
Variable Total Paper Cont.

# of Observations 198 193 196
Log likelihood value 10.46 -136.15 -130.12
House? 0.31 ** 0.71 ** 1.38 **
 (0.10) (0.24) (0.40)
2nd floor 0.0233 0.18 * 0.12
 (0.048) (0.11) (0.13)
3rd floor 0.042 0.17 0.014
 (0.050) (0.11) (0.12)
4th floor 0.11 * 0.34 ** 0.46 **
 (0.065) (0.15) (0.19)
Distance to bins -0.019 -0.0929 -0.24 **
 (0.026) (0.060) (0.088)
Distance to -0.041 -0.25 0.13
 bins x House? -0.12 (0.29) (0.30)
Enough space? 0.16 ** 0.18 ** 0.42 **
 (0.037) (0.084) (0.094)
Enough space? -0.25 ** -0.39 ** -0.81 **
 x House? (0.082) (0.19) (0.23)
Recycle away 0.046 0.20 ** 0.16 *
 from home? (0.035) (0.079) (0.081)
Newspaper 0.073 * 0.11 -0.012
 subscriber? (0.038) (0.087) (0.095)
Less than full time? -0.0016 0.046 -0.049
 (0.045) (0.10) (0.12)
More than full time? -0.00088 -0.0011 -0.059
 (0.033) (0.075) (0.081)
2 adults in 0.041 -0.0072 0.19 *
 household? (0.038) (0.086) (0.11)
> 2 adults in -0.027 -0.060 -0.024
 household? (0.059) (0.13) (0.17)
Answered age? -0.19 ** -0.16 -0.22
 (0.066) (0.15) (0.17)
Age of oldest 0.004 ** 0.0031 0.0034
 person (0.0013) (0.0028) (0.0032)
Education of 0.0077 ** 0.013 -0.0082
 respondent (0.0039) (0.0088) (0.019)
All male? 0.11 ** -0.14 0.15
 (0.0548) (0.11) (0.13)
All female? 0.15 ** 0.092 0.32 **
 (0.042) (0.093) -0.13
Building dummies not sig. not sig. Sig.
Standard error 0.20 0.44 0.44
 (ancillary) -(0.011) (0.028) (0.031)

 MFDs Only

Sample Dependent
Variable Total Paper Cont.

# of Observations 135 131 133
Log likelihood value 3.24 -100.18 -89.87
House?

2nd floor 0.039 0.22 * 0.13
 (0.054) (0.12) (0.13)
3rd floor 0.0037 0.19 0.025
 (0.054) (0.12) (0.13)
4th floor 0.18 ** 0.40 ** 0.44 **
 (0.085) (0.17) (0.20)
Distance to bins -0.034 -0.040 -0.26 **
 (0.037) (0.068) -(0.091)
Distance to
 bins x House?
Enough space? 0.17 ** 0.19 ** 0.43 **
 (0.040) (0.092) (0.097)
Enough space?
 x House?
Recycle away 0.051 0.19 ** 0.093
 from home? (0.043) (0.11) (0.10)
Newspaper 0.080 0.12 0.040
 subscriber? (0.053) (0.13) (0.13)
Less than full time? 0.071 0.34 ** 0.18
 (0.070) (0.15) (0.17)
More than full time? 0.0071 0.036 -0.017
 (0.040) (0.096) (0.094)
2 adults in 0.044 0.024 0.41 **
 household? (0.065) (0.11) (0.16)
> 2 adults in -0.064 -0.13 0.20
 household? (0.10) (0.18) (0.26)
Answered age? -0.19 ** -0.17 -0.30
 (0.088) (0.20) (0.22)
Age of oldest 0.0039 * 0.0024 0.00024
 person (0.0021) (0.0048) (0.0049)
Education of 0.0025 0.012 0.022
 respondent (0.011) (0.011) (0.026)
All male? 0.14 ** -0.18 0.23
 (0.068) (0.13) (0.16)
All female? 0.17 ** 0.064 0.44 **
 (0.067) (0.12) (0.17)
Building dummies Sig. not sig. Sig.
Standard error 0.20 0.47 0.44
 (ancillary) (0.014) (0.039) (0.037)

Notes: Standard errors are in parentheses. * indicates significant
at 10% level; ** indicates significant at 5% level. "Sig" means
that at least 1 of the 13 dummies was significant or the set was
jointly significant. "Not sig." means that no single dummy was
significant and the set was jointly insignificant. Joint
significance was determined with a likelihood-ratio test.
Where the dummies were insignificant, the results presented
are for the regression with the building dummies dropped.

TABLE 3
Results of Probit Analyses of MFD Recycling Participation

Dependent variahle Total Paper

# of Observations 135 131
Log-likelihood value -41.58 -51.675
Pseudo [r.sup.2] 0.24 0.21
2nd floor -0.11 (0.084) 0.053 (0.078)
3rd floor -0.12 (0.089) 0.014 (0.083)
4th floor 0.038 (0.056) 0.17 ** (0.043)
Distance to bins -0.024 (0.033) 0.017 (0.052)
Enough space? 0.099 ** (0.048) 0.12 * (0.068)
Recycle away from home? 0.080 (0.060) 0.18 ** (0.095)
Newspaper subscriber? 0.074 (0.037) -0.038 (0.094)
Less than full time? 0.062 (0.039) 0.16 * (0.046)
More than full time? -0.039 (0.048) -0.039 (0.071)
# of adults in household 0.14 ** (0.051) 0.021 (0.066)
Education of respondent 0.014 (0.012) 0.0054 (0.019)
All male? 0.066 (0.051) -0.22 ** (0.13)
All female? 0.093 (0.057) -0.059 (0.11)

Dependent variahle Cont.

# of Observations 133
Log-likelihood value -39.67
Pseudo [r.sup.2] 0.33
2nd floor -0.014 (0.048)
3rd floor -0.083 (0.069)
4th floor 0.033 (0.038)
Distance to bins -0.036 (0.029)
Enough space? 0.12 ** (0.055)
Recycle away from home? 0.041 (0.044)
Newspaper subscriber? 0.039 (0.033)
Less than full time? 0.062 * (0.031)
More than full time? -0.051 (0.043)
# of adults in household 0.19 ** (0.060)
Education of respondent 0.021 ** (0.011)
All male? 0.12 ** (0.049)
All female? 0.11 ** (0.053)

Notes. Marginal effects (dF/dx) are reported instead of
coefficient values. Standard errors are in parentheses.
* indicates significant at 10% level:
** indicates significant at 5% level.

TABLE 4
Effects of Independent Variables on MFD Recycling Rates

 Total Paper

Elasticity of recycling rate
 (dln[rate]/dln[X]) with
 respect to (a)
 Distance to bins -0.17 (0.19) -0.12 (0.21)
 The presence of adequate space 0.28 ** (0.071) 0.20 ** (0.10)
Marginal effect on recycling rate
 (d[rate]/d[X]) of (a)
 Distance to bins -0.034 (0.037) -0.040 (0.068)
 The presence of adequate
 space (b) 0.17 ** (0.040) 0.19 ** (0.092)

 Cont.

Elasticity of recycling rate
 (dln[rate]/dln[X]) with
 respect to (a)
 Distance to bins -0.66 ** (0.24)
 The presence of adequate space 0.37 ** (0.086)
Marginal effect on recycling rate
 (d[rate]/d[X]) of (a)
 Distance to bins -0.26 ** (0.091)
 The presence of adequate
 space (b) 0.43 ** (0.097)

(a) These effects are calculated by applying STATA's marginal-effects
routine to the results of the tobit regressions reported in
Table 2. Calculations are done using variable means. Standard
errors are in parentheses. * indicates significant
at 10% level; ** indicates significant at 5% level.

(b) These calculations give the effect of changing this
dummy variable from 0 to 1.


ABBREVIATIONS

EPA: Environmental Protection Agency MFD: Multifamily Dwelling SFD: Single-Family Dwelling

(1.) See Fullerton and Kinnaman (1996), Hong and Adams (1999), Jenkins et al. (2000), Kinnaman and Fullerton (2000), Reschovsky and Stone (1994), Sterner and Bartelings (1999), and Van Houtven and Morris (1999) for examples.

(2.) Here we describe the recycling program in Urbana as it was when we collected data. The program has changed since then. For example, SFDs no longer must sort their recycling.

(3.) One difference is that the MFD recycling program accepts a slightly broader range of containers, including dairy/juice cartons and more types of plastic bottles.

(4.) We did not include a question in the survey to elicit income, for fear that this would seriously reduce the response rate. The building-level fixed effects we use may help control for variation in income, because income is likely to be correlated with rent.

(5.) In the case of SFDs, we assumed that residents would put their recycling bins on the point of the curb closest to the entrance to the house.

(6.) We carefully controlled for variation in stride length among the three survey distributors.

(7.) The terms building and complex are used interchangeably. For the fixed effects, units in different buildings are considered part of the same complex if they share recycling bins.

(8.) There is one senior-citizen apartment building that has eight stories, but it is a secured building that we could not access for our survey.

(9.) The mean number of years of education is still remarkably high in our sample--16.6 years in the combined sample. This is, however, consistent with the experience reported by Ebreo and Vining (2000, p. 159) after a decade of survey research in the same community.

(10.) Indeed, although the logit routine in Limdep was able to run on these data with a large number of the limit observations removed, it would not converge when applied to the full data sets.

(11.) In all three of the first columns, F-tests fail to reject the hypothesis that the coefficient on "Enough space?" added to the coefficient on "Enough space? x House?" sums to zero.

(12.) We thought that many apartment dwellers would be unlikely to be able to give an accurate estimate of the number of square feet in their homes and did not want to lower response rates by asking a question for which respondents would need to research the answer.

(13.) It is also hypothetically possible that the causality runs in reverse and that households with a strong interest in recycling are choosing dwellings that have adequate interior storage space, but it seems unlikely that recycling convenience is a major factor in households' housing choices.

(14.) In Syracuse, New York, for example, some high-rise residents have recyclables picked up at their door, but must dispose of trash in a more distant chute (EPA 1999).

REFERENCES

Ebreo, A., and J. Vining. "Motives as Predictors of the Public's Attitudes toward Solid Waste Issues." Environmental Management, 25(2), 2000, 153-68.

Fullerton, D., and T. C. Kinnaman. "Household Responses to Pricing Garbage by the Bag." American Economic Review, 86(4), 1996, 971-84.

Greene, W. H. Econometric Analysis, 4th ed. Upper Saddle River, NJ: Prentice Hall, 2000.

Hong, S., and R. M Adams. "Household Responses to Price Incentives for Recycling: Some Further Evidence." Land Economics, 75(4), 1999, 505-14.

Jakus, P. M., K. H. Tiller, and W. M. Park. "Generation of Recyclables by Rural Households." Journal of Agricultural and Resource Economics, 21 (1), 1996, 96-108.

--. "Explaining Rural Household Participation in Recycling." Journal of Agricultural and Applied Economics, 29(1), 1997, 141-48.

Jenkins, R. R., A. M. Salvador, K. Palmer, and M. J. Podolsky. "The Determinants of Household Recycling: A Material-Specific Analysis of Recycling Program Features and Unit Pricing." Journal of Environmental Eeonomics and Management, 45(2), 2003, 294-318.

Kinnaman, T. C., and D. Fullerton. "Garbage and Recycling with Endogenous Local Policy." Journal of Urban Economics, 48(3), 2000, 419-42.

McQuaid, R. W., and A. R. Murdoch. "Recycling Policy in Areas of Low Income and Multi-Storey Housing." Journal of Environmental Planning and Management, 39(4), 1996, 545-62.

Morris, G. E., and D. M. Holthausen Jr. "The Economics of Household Solid Waste Generation and Disposal." Journal of Environmental Economies and Management, 26(3), 1994, 215-34.

Reschovsky, D., and S. E. Stone. "Market Incentives to Encourage Household Waste Recycling: Paying for What You Throw Away." Journal of Policy Analysis" and Management, 13(1), 1994, 120-39.

Rushforth, C. U-Cycle Data Reports. 10 December 2003. Email.

Sterner, T., and H. Bartelings. "'Household Waste Management in a Swedish Municipality: Determinants of Waste Disposal, Recycling and Composting." Environmental and Resource Economics, 13(4), 1999, 473-91.

Stevens, B. J. Multi-Family Recycling: Costs, Diversion, and Program Characteristics. Washington, DC: U.S. Conference of Mayors, 1999.

U-Cycle. Interesting Recycling Facts. online document available at www.city.urbana.il.us/urbana/public_works/ env-recyclefacts.html, 4/01, 1995.

U.S. Census Bureau. Demographic Profile, Urbana City. Illinois. Online document available at http:// censtats.census.gov/data/il/1601777005.pdf, 3/04, 2000.

U.S. Environmental Protection Agency. Characterization of MSW in the U.S.: 1998 Update. Washington, DC: Environmental Protection Agency, 1998.

--. Complex Recycling Issues: Strategies for Record-Setting Waste Reduction in Multi-Family Dwellings. EPA-530-F-99-022, available online at www. epa.gov/osw, 1999.

--. Multifamily Recycling. Washington, DC: U.S. Government Printing Office, 2001.

Van Houtven, G. L., and G. E. Morris. "Household Behavior under Alternative Pay-as-You-Throw Systems for Solid Waste Disposal." Land Economics, 75(4), 1999, 515-37.

Vining, J., and A. Ebreo. "What Makes a Recycler? A Comparison of Recyclers and Nonrecyclers." Environment and Behavior, 22(1), 1990, 55-73.

Westergard, D. L. "A Cost-Benefit Analysis of a Proposed Multifamily Curbside Recycling Program in Champaign-Urbana, Illinois." M.S. thesis, University of Illinois at Urbana-Champaign, 1996.

AMY W. ANDO and ANNE Y. GOSSELIN *

* For helpful comments and suggestions, we are indebted to Dennis Jansen and two anonymous referees, John Braden, Don Fullerton, Robin Jenkins, and Eric Rasmusen: attendees of presentations at Cambridge University; the University of Illinois, and Indiana University: the Heartland Environmental and Resource Economics Workshop; and conferences of the EAERE, SEA, and WEA. This material is based in part work supported by the Cooperative State Research, Education and Extension Service, U.S. Department of Agriculture, under project no. ILLU 05-0305. We also gratefully acknowledge support from the Jonathan Baldwin Turner Undergraduate Research Program.

Ando: Assistant Professor, Dept of Agricultural and Consumer Economics, University of Illinois at Urbana-Champaign, 326 Mumford Hall, 1301 W. Gregory Drive, Urbana, IL 61801. Phone 1-217-333-5130, Fax 1-217-333-5538, Email amyando@uiuc.edu

Gosselin: Graduate Student, Dept. of Agricultural and Consumer Economics, University of Illinois at Urbana-Champaign, Urbana, IL 61801. Phone 1-217-333-7683, Fax 1-217-333-5538, Email agosseli@ uiuc.edu
COPYRIGHT 2005 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2005 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Ando, Amy W.; Gosselin, Anne Y.
Publication:Economic Inquiry
Geographic Code:1USA
Date:Apr 1, 2005
Words:8672
Previous Article:Information asymmetry and competitive bidding in auditing.
Next Article:Determining output and inflation variability: are the Phillips curve and the monetary policy reaction function responsible?
Topics:

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters