Determinants of Per Capita Municipal Solid Waste Generation in the Southeastern U.S.

running title: Per Capita Solid Waste Generation

Daniel Hockett Douglas Lober*

Keith Pilgrim

Duke University School of the Environment

P.O. Box 90328 Durham, NC 27706

9 19-6 13 -8029 9 19-684-874 1 (FAX)

A u p t 4, 1994

* address all correspondence to this author

Abstract

This study's goal is to identify and measure the variables which influence per capita municipal solid waste generation in the southern U.S., using the 100 counties of North Carolina as a data set and regression models for statistical analysis. Variables are selected to capture the residential, institutional, and commercial components of the municipal solid waste stream as well as the overall structure of waste management through inclusion of waste disposal fees. An additional goal of this study is to examine the influence of the components of retail sales, including sales of eateries, merchandise, food stores, and apparel stores, as the retail industries have been suggested to contribute significantly to waste generation.

The results indicate that retail sales and the waste disposal fee are the significant determinants of waste generation while variables to account for manufacturing, construction, personal income, and degree of urbanization did not prove significant. Of the components of retail sales, per capita sales of eating establishments proved the greatest influence on waste generation. Policy implications are that an increasing effort should be made to focus on eating establishments and to establish appropriate tipping fees if a goal of waste reduction is to be achieved.

1. Introduction

The U.S. currently generates and disposes of almost 200 million tons per year of

paper, plastics, yard waste, glass and other materials, which are collectively labeled

municipal solid waste (MSW) or more commonly, garbage (U.S. EPA, 1992). This waste

is primarily produced by residential, institutional, and commercial sources. Municipal solid

waste is regulated under subtitle D of the main law covering waste management, the

Resource Conservation and Recovery Act (RCRA). Municipal solid waste is distinguished

from other subtitle D wastes such as the 7.6 billion tons of industrial waste and the 5.2

billion tons of mining, oil and gas waste generated annually. It is also separate from the

200 million tons generated per year of hazardous waste which is classified as a subtitle C

waste under RCRA.

When municipal solid waste generation data is calculated on a per capita basis,

each individual in the U.S. is said to generate around .75 tons per year or 4 pounds per

person per day (U.S. EPA, 1992). However, there is considerable variation in waste

generation across regions in the U. S. Chattanooga, Tennessee, for example, generates 9.4

pounds per person per day while Yakima, Washington produces 1.9 pounds per person

per day (U.S. Congress, 1989). A number of variables, such as local climate, the economy,

demographic characteristics of the population, the amount of tourism in the region, and

population density are potential variables which may account for the variation in amount

and type of waste produced when waste is measured on a per capita basis.

This paper's goal is to identifl and measure variables which influence per capita

MSW generation, using the counties of North Carolina as a data set. The paper first

describes the nature of the economy, landscape, and waste generation in North Carolina.

It then briefly reviews the literature concerning the determinants of waste generation.

Next, a model to characterize the influence of a number of variables on waste generation is

created. The results indicating significant determinants of waste generation are presented

and policy implications are discussed.

2. The North Carolina Wasteshed

2.1 North Carolina landscape, economy and population

North Carolina, with a population of 6,628,000 people, is an extremely diverse

state in terms of landscape, economy, and population. The State has four main geographic

regions: Coastal, Coastal Plain, Piedmont, and Mountain. The Coastal region, with 10

percent of the population, includes the Cape Hattaras National Seashore. The Piedmont

region, with 55 percent of the population, has three major population centers, the Raleigh-

Durham-Research Triangle area, the Greensboro-Winston-Salem Triad area, and

Charlotte, as well as the Interstate 40 corridor, one of the fastest growing business

locations in the United States. The Mountain region, with 14 percent offhe population,

includes both the Blue Ridge Parkway and Great Smoky Mountains National Park, the

most visited national park in the United States. In fact, North Carolina is a major tourist

destination with sixty-one million people visiting tourist sites in North Carolina in 1988

(North Carolina Department of Travel and Tourism, 1988 ).

Approximately 75 percent of the land area in North Carolina is classified as

farmland or forest land. Fifty seven percent of the state lives in urban areas where urban is

defined as any county classified as a federally-designated Metropolitan Statistical Area

(MSA). The Piedmont area is particularly urbanized with 8 1 percent of its population

2

' . !

living in urban counties. Fifty seven percent of the waste is disposed of in the urban

counties of the state (N.C. Department of Environment, Health, and Natural Resources,

1992).

The largest employers for North Carolina's 3,713,000 workers are the government,

manufacturing companies, service industries and retail operations (N. C. Department of

Environment, Health, and Natural Resources). In all four regions of the state, retail

employers account for between 16 and 19 percent of the total employment. Manufacturing

industries employ 11 percent of the workers in the Coastal Region and 22 to 25 percent in

the rest of the state. The service industry accounts for between 15 and 2 1 percent of the

workers in all four regions. The government employs 33 percent of the employees in the

Coastal Region, 23 percent in the Coastal Plain, 12 percent in the Piedmont, and 16

percent in the Mountain Region.

2.2 North Carolina waste generation

In 1990, North Carolina's counties disposed of 6,823,000 tons of municipal solid

waste. Ninety-eight percent of this was disposed in landfills, primarily at the county level.

The counties also recycled 436,000 tons of material and composted 267,000 tons of yard

waste for a state-wide recycling and composting rate of 9.2 percent (N.C. Department of

Environment, Health and Natural Resources, 1993). Given the state's population of

6,628,000 people, the per capita waste generation was 1.1 tons per person per year or 6.2

pounds per person per day. Fifteen counties accounted for over 50 percent of the waste

disposed, with Mecklenburg County, the location of the State's largest city, Charlotte,

contributing the greatest amount of waste disposed, 8.8 percent of the total. All but six of

North Carolina's counties had landfills receiving municipal solid waste during the study

period.

These state-wide figures obscure considerable variability in waste generation

among counties (N. C. Department of Environment, Health, and Natural Resources,

1994). For example, Dare County, which contains the Cape Hatteras National Seashore

and therefore has a high tourist component, disposed of 2.2 tons per person per year or 12

pounds per person per day. Wilson County, which did not have a tipping fee when the

data were collected, produced 1.8 tons per person per year or nearly 10 pounds per capita

per day. Caswell and Camden counties produce .25 and .3 1 tons of MSW per person per

year.

Figure 1 cartographically represents the per capita waste generation in North

Carolina. No clear pattern exists to account for the differences in waste generation among

counties. This raises the main research question of this study: what can account for the

difference in waste generated at the county level, when controlling for population

differences?

[figure 1 3

3. Determinants of waste generation

The overriding variable influencing total waste generation is population. The

Piedmont region contains 55 percent of the population and disposes of 59 percent of the

State's waste. The coastal plain has 21 percent of the population and disposes of 19

percent of the waste. The Coastal region, with 10 percent of the population, disposes of

10 percent of the waste. The Mountain region disposes of 13 percent of the waste and

4

contains 14 percent of the population. Figure 2a shows this strong correlation for the

counties of North Carolina between the population in a county and the amount of waste

produced.

As would logically follow from figure 2a, when the amount of waste produced is

presented on a per-capita basis as in figure 2b, there is no relationship between population

and waste generation More populous counties simply do not produce more waste per

person than do less populated counties.

[figure 2a and figure 2b]

A variety of studies have attempted to account for differences in per capita waste

generation although this remains an area needing hrther research (Vesilind, et al, 1977).

We briefly mention some of the findings, as reviewed by Ali Khan and Burney (1 989),

Henricks (1 984), and ourselves, with an emphasis on the variables included in our study

to provide context and support for our hypotheses, which are presented in section 5.

AIncome and affluence have often been found to be positively correlated with

waste generation (Chang et al, 1993; Dayal et al, 1993; Jenkins, 1993; Rhyner, 1976;

Richardson and Havlicek, 1978; Wertz, 1976) although there are null findings (Ali Khan

and Burney, 1989, Cailas et al, 1994; Henricks, 1994) and negative ones (Grossman,

1974; Rathje and Murphy, 1992) as well. Cailas et a1 (1994) and Jenkins (1993)

determined that a negative correlation exists between population density and waste

generation while Ali Khan et a1 (1989) found no correlation. Studies by Cailas et a1 (1994)

and Henricks (1 994) showed a positive correlation between degree of urbanization and

waste generation while Rhyner's (1 976) examination found none. Dayal et a1 (1 993) and

5

,

Jenkins (1993) discovered a correlation between climate and waste generation while Ali

Khan et a1 (1989) found no effect. Jenkins (1993) found that there was a positive

correlation between waste generation and age while Richardson and Havlicek (1 978)

found that those who were middle-age rather than young or old produced more waste.

Jenluns( 1993) indicated that smaller household sizes produced more waste per capita

while Cailas et a1 (1 994) and Rhyner (1 976) found no effect. These studies analyze a

variety of localities and regions from cities in developing countries to several states. Only

the study by Rhyner (1 976) applies to North Carolina and the surrounding region.

Correctly characterizing the causal mechanisms for these results is often difficult.

For example, wealthier people often use and discard more paper, particularly as they read

more newspapers. However, they produce less cans as they eat fewer canned foods. As

paper often constitutes a greater proportion of the average waste stream, the net effect

may be a positive correlation between income and waste generation. Nonetheless, the net

effect of income on waste generation can be ambiguous. One method to improve this

analysis is to characterize waste generation by material such as glass, paper, plastic, and

yard waste, as has been done in some studies (Henricks, 1994; Richardson and Havlicek,

1978; Ali Khan et al, 1989 )

Generally the models of per capita waste generation that exist have several

shortcomings. One, they often focus exclusively on demographic variables and therefore

measure primarily residential sources of waste, despite the importance of commercial and

institutional generators to the waste stream. For example, such contributors to the waste

stream as tourists would not be captured by variables to measure residential waste. Several

researchers have tried to incorporate indicators of commercial waste into waste generation

models. Henricks (1994) found that per capita retail sales was positively correlated with

6

total waste production in Florida. Gay et a1 (1993) develop a predictive model for waste

generation based on retail sales.

A second shortcoming of waste generation studies is that the models exclude key

structural variables, such as waste disposal fees, which may influence the amount of waste

disposed. The result of this exclusion is that the relative importance of all variables

influencing waste generation cannot be determined.

Another shortcoming of previous studies is that the data that is used to create

models is often collected by different methods or is inconsistent in definition, making it

non-comparable. For example, one set of data may include construction and demolition

debris while another excludes it in the measure of total waste generation.

These shortcomings provide an opportunity to develop studies which better reflect

the scope of the municipal waste stream and key variables which influence it while using

data which is comparable, an approach which is followed here.

4. Goals and Method

The main goal of this study is to create a model of the demographic, economic,

and structural determinants of per capita waste generation for the southern region of the

U. S. Specifically, the intent is to explain waste generation by identifling both the

significant determinants of waste production and their relative importance, with less

emphasis on constructing a predictive model. The independent variables chosen for

inclusion are those which are explicitly related to waste disposal patterns and include

economic, structural, and demographic variables. Economic variables include per capita

7

retail sales, per capita value added by manufacturing and per capita construction costs. A

structural variable examined is the cost per ton to dispose of the waste at the landfill,

which is called the tipping fee. Demographic variables include per capita income and urban

population percentage.

A second goal of the study is to closely examine the specific nature of the

contribution of retail sales to the generation of waste. To do so, the study examines how

per capita sales of apparel stores, food stores, merchandise, and eateries contribute to

overall production of waste.

The county was chosen as the unit of analysis for this study due to data availability

and the political appropriateness of this size unit for waste management. Furthermore, the

county level provides significant variation in both the dependent and independent variables

of interest to allow for the testing of their relationship. The amount of waste generated,

tipping fees, and populations for the 100 North Carolina counties were obtained from the

199 I-92 North Carolina Solid Waste Management Annual Report (Department of

Environment, Health, and Natural Resources (NCDEHNR), 1993). Demographic

information was obtained from the Census Atlas of North Carolina. Economic data was

acquired from the Statistical Abstract of North Carolina Counties (1991) and from

Courtney et al's (1992) 1992 County and City Extra of North Carolina. Complete data

was available for 89 counties which is summarized in table 1 below.

Several regression models were constructed to determine the significance and

importance of variables of interest. A number of statistical tests were done to assure that

the assumptions necessary to use ordinary least squares regression were met. These tests

are presented in section 6 .

8

5. Hypotheses and Model Creation

5.1 The dependent variable: per capita waste generation

The dependent variable used in this study is the amount of waste generated by

weight per capita per day. This figure included municipal solid waste primarily from

residential, commercial, and institutional sources, but including some construction and

industrial waste. Each county determined and reported to the State the amount of waste

which was disposed of at its landfill as well as the tons of material recycled and the

amount of yard waste disposed so that a total waste generation figure could be calculated

by summing these amounts.

5.2 Independent variables

Many variables can be considered as possible determinants of waste generation in

the state. North Carolina's waste managers indicate that the most important factors

influencing the generation and disposal of waste are the "geography, population, economic

base, income, land use, and available transportation routes'' (NC Department of

Environment, Health, and Natural Resources, 1992, p. ES-3). The State waste managers

also conclude that residential waste is a hnction of population and commercial and

industrial waste depend on tourism and the type of industry in the region. Variables to

capture all these potential sources of waste were included in this study along with others

suggested by the literature. In addition, the variables were selected so that they could be

easily reproducible in examinations of other regions. U.S. Bureau of the Census Standard

Industry Codes (SIC) provided the basis for the data on sales for different sectors of the

9

economy. The variables were measured on a per capita basis where appropriate to make

them consistent with the dependent variable of interest. Table 1 identifies the independent

variables of interest for this study as well as the units of measurement and the source of

the data.

[Table 13

The type of economic activity of the state is hypothesized to influence the

generation of waste. This economic activity was partially captured by including variables

to measure retail sales, construction activity, and manufacturing. It should be noted that

retail sales is also a measure of tourism. Tourists spend considerable amounts of money on

food, clothing, and merchandise which generate packaging, shipping, food, and other

waste. Value added to manufactured goods by companies in the county was the best

known available surrogate measurement for general industrial activity which can

contribute packaging or scrap to waste generation. Construction costs were included in

the analysis to account for possible construction or demolition debris deposited in the

landfill.

It is hypothesized that the degree of urbanization effects the waste generated.

Urban population was considered as an independent variable because this sector of the

population is believed to rely heavily on standard municipal solid waste pickup versus

alternative methods of disposal such as backyard disposal or burning.

The size of the tipping fee has the potential to influence waste disposal as greater

tipping fees might lead to less waste generated, alternative disposal methods or efforts to

find cheaper disposal sites, including out of the county. The average tipping fee in the

10

State was $16.06/ton with a range from $60 to $6 per ton for those counties who charged

for waste disposal. Twenty-six counties did not charge to dispose of waste.

The full model to be tested takes the following form, as presented in equation 1 :

WASTE=bo + blTIPFEE + b2RETSAL + b3VALUAD + b4CONST + b5lNCOM +

b6URBPOP (1)

6. Data analysis

Analyses were performed to ensure that the independent variables included in the

model, per capita retail sales, value added manufacturing, per capita construction costs,

and per capita income did not violate the regression assumptions of a linear relationship

between independent and dependent variables, independence of the independent variables,

and constant variance and normality of errors.

First, potential independent variables were analyzed for correlation. Variables that

were highly correlated with another explanatory variable were not considered independent.

A decision was then made to exclude one of the correlated variables based on which

variable alone was the best predictor of per capita waste generation.

Score tests were utilized to check for non-constant variance of the errors. A

general Score test was first performed by regressing the scaled squared residuals (Ui’s)

against the fitted values of the large model. The results of this test demonstrated that the

errors exhibited constant variance. Score tests were also pedormed to investigate the

influence of the independent variables on the variance of the residuals. These tests were

11

conducted by regressing the Ui'S against each independent variable. Results of these tests

demonstrated that the variance of the residuals were not a function of any of the

descriptors. Plots of the residuals versus each independent variable did not exhibit any

non-linear trends. A plot of the standard normal quartiles versus the residuals

demonstrated a strong linear trend indicative of normal distribution of the residual errors.

Case diagnostics from this model revealed that none of the counties were outliers nor did

any bear disproportionate influence on the model.

In summary, the assumptions for ordinary least squares regression with no

necessary transformations were met for the model of waste generation against per capita

income, per capita retail sales, per capita value added by manufacturing, per capita

construction costs, and tipping fee.

7. Results and discussion

7.1 Significant variables

The model with parameter estimates and standard errors in parentheses is as

follows:

WASTE= 3.725 -.034*TIPFEE + .323* RETSAL + .059*VALUAD + .227*CONST - (.859) (.013) (. 069) (. 043) (.441)

.OOO* INCOM + .007*URBPOP (.OOO) (.009)

Two variables, per capita retail sales (p = . O O O i ) and tipping fees (p = .0104),

proved to be significant determinants of waste generation. As retail sales increased, so did 12

the amount of per capita waste generation. Higher tipping fees were associated with lower

levels of total waste levels.

According to our model, a $1000 increase in per capita retail sales will result in a

.323 pound per day increase in per capita waste generation. This finding of the importance

of retail sales is consistent with the results of Gay et a1 (1993) who found retail sales to be

an excellent predictor of waste generation.

Since per capita retail sales was the single best predictor of per capita waste

disposal, we decided to hrther investigate the nature of the effect of retail sales on waste

disposal. Data for retail sales receipts was divided into four major sub-classes: eating

establishments, food stores, general merchandise stores, and apparel stores (Courtney, et

al, 1992). A separate model, equation 3, was then constructed with these four sub-classes

used as independent variables.

WASTE= bo + blCLOTHES + b2EATS + b3MERCH + b4FOOD (3)

Equation 4 presents the parameter estimates for this model, with standard errors in

parentheses. These results illustrate that per capita sales of eating establishments produced

the only significantly non-zero slope, demonstrating the significant contribution of

restaurants to the waste stream.

WASTE=3.47 - .000"CLOTHES + ,002"EATS + .001*MERCH + 001"FOOD (4)

(.581) (.002) (. 002) (.001) (.001)

13

r2 = .371

Tipping fee is the only significant variable with a negative beta, indicating that an

increase in tip fee is associated with a decrease in waste both deposited at the landfill and

recycled. A dollar increase in tipping fee is estimated to result in a .034 pound per day

decrease in per capita waste generation. This finding of the significance of tipping fees is

relevant to waste management for several reasons. One, it strongly indicates the value of

having a tipping fee to control waste generation as those counties without tipping fees had

higher rates of waste disposal. Two, it indicates the relatively small importance of

demographic variables relative to structural ones in determining waste generation. What

remains unexplained by our finding is what happened to the waste when tipping fees were

higher. Was there less waste generated or was it disposed of illegally? We do know that

11 counties received waste from other counties (N.C. Department of Environment,

Health, and Natural Resources, 1993). However, the county waste generation data has

been adjusted for this. Research on unit based pricing, which involves charging individuals

by the amount of waste generated, has indicated that this can lead to 10 to 30 percent

declines in the amount of waste generated (Miranda, forthcoming). It has not yet been

determined exactly how the waste is reduced. Some mechanisms include changes in

consumer purchasing behavior and illegal disposal.

Determining the relative importance of these two variables, per capita retail sales

and tipping fees, is difficult given the different scales used to measure each. One such

method is to examine the sum of the squares for each variable. Since per capita retail sales

has a higher sum of the squares, 33.44, than does the tipping fee, 10.77, it can be

considered more important using the sum of the squares criterion. Another approach to

determining relative importance is to examine the standardized estimates of the two

14

variables. These show that as one moves from the lowest to highest value within the

observed range of the retail sales variable, the effect on waste generation is greater relative

to the effect of the tipping fee variable within its observed range.

7.2 Non-significant variables

Interestingly, the slope estimates for income, urbanization, manufacturing, and

construction were not significantly different from zero, indicating that these three variables

do not explain variability in per capita waste generation. The statistically insignificant beta

for construction costs might suggest that construction debris is either not deposited in

municipal solid waste landfills or is not significant (Apotheker, 1990). However, in

general, construction and demolition debris in North Carolina is deposited in municipal

solid waste landfills. Land clearing and inert debris, on the other hand, is put in special

demolition landfills and is not measured in the waste generation data. This would lead to

the expectation that a variable measuring this type of waste would not be significant as a

determinant of total waste generation as measured in North Carolina. Our null finding

concerning income and degree of urbanization is consistent with the rather ambiguous

findings in the literature.

7.3 Model uncertainty and error

There are a number of sources of both uncertainty and error which can enter into

our model. These relate to the accuracy and precision of the data which are used for

parameter estimation as well as the adequacy of the measures which are used to represent

this waste generation.

15

One source of possible error is that there is likely to be considerable inaccuracy in

the measuring and reporting of waste by county. Scales to weigh garbage may not be

operational, waste may enter landfills without being weighed, personnel may not be

trained adequately in waste measurement, waste may be illegally disposed of or

transported out of state, and several landfills opened during the course of the year where

the data for this study was collected. The recycling rate is probably underreported as well

as some commercial establishments, such as supermarkets, may have their own non-

reported recycling programs for materials such as corrugated cardboard.

A second source of potential error is that the per capita waste generation measures

by county may not be comparable as they might include different types of waste. Some

may include industrial or construction and demolition waste while others do not,

depending on both the training of those reporting and the range of disposal options

available to a county.

A third source of possible error is that the included variables may not be adequate

measures of the underlying economic or structural process which they seek to measure.

The finding, for example, that per capita construction costs was not significant does not

necessarily mean that construction and demolition waste is not a contributor to overall

waste generation but that this measure may not adequately capture it.

A fourth source of potential error is that the effect of time may not be adequately

considered (Chang et al, 1993). One such example of this is that some of the measures of

commercial activity are from different time periods than those concerning the amounts of

waste generated.

16

The purpose of this analysis was to identifl significant factors that explained

variability in per capita waste generation and to examine their relative importance rather

than predict per capita waste generation. A better predictive model of this complicated

process would require identification of other variables that are directly related to waste

production. Examples of such excluded variables are those to capture the government and

service sectors of the economy which can be significant contributors to waste generation.

These possible sources of uncertainty and error should be addressed in hture

research. Field studies, for example, can better characterize errors associated with waste

generation data or excluded waste streams. The model needs to be re-examined in hture

years as data for waste generation for each new year becomes available. New measures to

reflect different waste generation activities need to be developed. Reducing possible

uncertainty and error will contribute to making our findings more robust.

7.4 The model and waste management

The model constructed attempts to describe and explain the production of waste in

North Carolina. Clearly the size of the overall population will be the best predictor for the

total amount of waste generated in a region. This model's value for waste managers is to

identifl key determinants of waste generation on a per capita basis to improve waste

management strategy.

One of our findings, the importance of structural characteristics of waste

management relative to economic and demographic variables, is important given the waste

management changes currently occurring in North Carolina as well as nationally. A

structural component in the solid waste disposal scheme which could significantly effect

17

the generation and disposal of waste as characterized by this study is the evolution of

waste management activities from the county level to a regional level, corresponding with

the creation of large regional landfills. In fact, North Carolina has recently given approval

for eight regional landfills. Though only one landfill in the state is currently attracting

significant amounts of out-of-state waste, this could change with the construction of the

proposed landfills, given the low tipping fees in the region. The result is that this structural

component could have far more impact on waste disposal and generation than any of the

variables included in the study. Waste managers must account for these potential impacts

in permitting these disposal changes.

A second finding is the possible importance of restaurants as contributors to the

solid waste stream. This suggests that waste reduction efforts should target restaurants to

significantly impact the waste stream. However, it is also possible that the high correlation

between waste generation and restaurant sales may reflect a different underlying economic

process. The significance of restaurant sales may be indicative simply of a high

concentration of all types of waste generators and activities, such as the presence of

institutions and commercial establishments. One way to try to distinguish between these

alternative explanations for waste generation is to do a waste stream analysis (Crissman,

personal communication). In fact, waste stream characterizations do exist for several of

the counties in North Carolina. These can be examined to see if the actual studies of the

type of waste generated seem consistent with what is suggested by our model, i.e. whether

high restaurant sales are contributing to high overall waste generation.

A third finding was that commercial sources other than the retail sector did not

prove to be significant contributors to waste generation. In fact, though our though our

manufacturing value added variable didn't capture it, there is some evidence that special

18

manufacturing waste streams may account for a considerable portion of the waste

disposed of in landfills (Manuel, 1994). Wilson County, for example, has significant

amounts of tobacco clippings from storage and processing warehouses. Other county

landfills might receive ash from power plants and municipal waste water treatment sludge.

Some of the waste disposed of also may also be due to one time commercial events such

as plant closings or major construction projects. These commercial sources and amounts

of waste are unlikely to be accounted for by our predictor variables. Field research would

help to better examine the waste stream to reveal if the manufacturing sector might require

special targeting. However, any waste management plan aimed at this sector of the

economy must recognize the rapid change occurring in the generation patterns of

companies as they adapt to new environmental concerns and regulations.

8. Conclusions

There is considerable variation in the amount of waste generated on a per capita

basis in the U.S. The model constructed is valuable to address this issue by providing

some empirical evidence as to the relative importance of a variety of variables in

influencing waste generation while using a data set that is relatively consistent in data

collection and reporting methods. Specifically, this study finds that per capita retail sales

and tipping fee are the significant determinants of per capita waste generation while other

variables, particularly demographic ones, prove to not be significant as correlates of waste

production. The finding of the importance of tipping fee is particularly valuable for it

stresses the relative importance of waste management structural characteristics in

influencing waste generation and disposal rather then only socioeconomic ones. The study

also suggests that the retail sector, including eating establishments, should be a prime

target for waste reduction efforts. These findings can be extended beyond the population

19

of the counties of North Carolina. Many of the states in the Southeastern U.S. and in the

U. S . Environmental Protection Agency's Region IV (including South Carolina, Georgia,

Tennessee and Kentucky) share similar economic and population characteristics to North

Carolina and it is likely that the results are relevant to these states as well.

Waste generation, and specifically over generation, rather than problems related to

waste disposal, is increasingly recognized as the underlying cause of the waste problem.

Our study attempts to shed hrther light on this problem to better inform waste policy. Not

only do we identifl some key variables effecting waste generation but we point out the

sensitivity of any waste management plan not only to demographics and economy but to

structural variables influencing the overall management of waste. The challenge for waste

managers will be to link these findings to waste reduction and management efforts.

20

Acknowledgments

We would like to thank Paul Crissman, Andrea Borresen, and Susan Wright of the Division of Solid Waste Management of the North Carolina Department of Environment, Health, and Natural Resources.

List of tables and figures

Figure 1 : Per capita waste generation by county in North Carolina

Figure 2a: Waste generation versus population by county in North Carolina

Figure 2b: Per capita waste generation versus per capita population by county in North Carolina

Table 1 : Descriptive statistics for determinants of per capita waste generation

Table 1. Hypothesized determinants of per capita waste generation in North Carolina

Description

per capita MSW disposal tipping fee

per capita retail sales

value added by manufacturing

per capita construction costs per capita sales of eateries per capita sales of merchandise per capita sales of food stores per capita sales of apparel stores per capita income

urban population percentage

Units Year Variable Source Mean name value

Ibs/person/day 1991- WASTE NCDEH 5.3 1992 NR report

1992 $/ton 1991- TIPFEE 16.6

$1000/person 1990 RETSAL NC 6.8 Statistical Abstract

and City Extra

1987 VALUADD County 5.2 11

1990 CONST 0.6

$/person 1987 EATS 419

1987 MERCH 3 96

1987 FOOD 1136

1987 CLOTHES 207

1990 INCOM Census 13345

<<

( 1

<<

<<

16

(1

11

16

11

(1

Atlas of NC

% 1990 URBPOP 26.7 l<

mum 1.4

0

1.8

0.09

0

0

0

3 19

0

909 1

0

- 12.2

60

19.8

20.5

2.69

2197

1068

3289

553

20040

87.8

References

Ali Kahn, M. and Burney, F. (1989) Forecasting solid waste composition. Resources, Conservation and Recycling. 3, 1- 17.

Apotheker, S. (1990) Construction and demolition debris-- the invisible waste stream. Resource Recycling 66-74.

1990 Census Atlas of North Carolina . Boone, NC: Appalachian State University

Cailas, M.D., Kerzee, R.G., Swager, R. and Anderson, R. 1993. Development and Application of a Comprehensive Approach for Estimating Solid Waste Generation in Illinois: First Phase Results. Urbana, ILL: The Center for Solid Waste management and Research University of Illinois Urbana-Champaign.

Chang, N., Pan, Y., and Huang, S. (1993) Time series forecasting of solid waste generation. Journal of Resource Management and Technology 21, 1-9.

Courtney, M., Flatter, Hall, G. F. (eds.) 1992 County and City Extra. Annual Metro City and County Databook. Landham, MD: Burana Press, .

Dayal, G. Yadav, A., Singh, R. P., and Upadhyay, R. 1993. Impact of Climatic Conditions and Socio-Economic Status on Solid Waste Characteristics: A Case Study. The Science of the Total Environment 136: 143-153.

Gay, A. E., Beam, R. G. and Mar, B. W. 1993. Cost-Effective Solid-Waste Characterization Methodology. Journal of Environmental Engineering 119, 63 1 - 644.

Grossman, D., Hudson, J. F., Marks, D.H. and Asce, M. (1974) Waste Generation Models for Solid Waste Collection. Journal of the Environmental Engineering Division. 100, 1219-1230.

Henricks, S. L. 1994. Socioeconomic Determinants of Solid Waste Generation and Composition in Florida. Master's thesis. Durham, NC: Duke University School of the Environment.

Manuel, J. 1994. Reaching the Goal in Our Future. The R-Word. Summer. North Carolina Recycling Association.

Miranda, M., Everett, J.W., Blume, D., and Roy, Jr. B.A. Market-Based Incentives and

Residential Municipal Solid Waste. Journal of Policy Analysis andManagement. forthcoming.

North Carolina Department of Environment, Health, and Natural Resources. 1992. North Carolina Recycling and Solid Waste Management Plan. Raleigh, NC: DEHNR.

North Carolina Department of Environment, Health, and Natural Resources. 1993. North Carolina Solid Waste Management Annual Report. 1991-1992. Raleigh, NC: NCDEHNR.

North Carolina Department of Travel and Tourism. 1988. 1988 North Carolina Travel Study. Raleigh, NC.

North Carolina Office of State Budget and Management, 1990 U.S. Census.

Rathje, W. and Murphy, C. (1992) Rubbish: The Archaeology of Garbage. NY: Harper Collins.

Richardson, R. A. and Havlicek, Jr., J. 1978. Economic Analysis of the Composition of Household Solid Wastes. Journal of Environmental Economics and Management 5 : 103-111.

Rhyner, C.R. 1976. Domestic Solid Waste and Household Characteristics. Waste Age. April. 29-39,50.

Statistical Abstract of North Carolina Counties for I991. 1991 State Data Center, Management and Information Services, Office of state Budget and Management. May.

U.S. Congress. Office of Technology Assessment. 1989. Facing America’s Trash: Wkat Next for Municipal Solid Waste. Washington D.C. : U. S. Government Printing Office.

U. S. Environmental Protection Agency. 1992. Characterization of Municipal Solid Waste in the US.: I990 Update. Washington, D.C.

Vesilind, P., Nissen, J. and McAlister, J. (1977) How to generate data for wastes allocation models. Solid Wastes Management 20, 68-72.

Wertz, K.L. 1976. Economic Factors Influencing Household’s Production of Refuse. Journal of Environmental Economics and Management, 2: 263 -272.

Figure 1. Per capita waste generation by county in North Carolina in 1991-92

Per Capita Municipal Solid Waste (tondperson)

0 ul c- L, N L, N c. 0

I 0

O O

0 0

0

I 1 1 1 1

0

0

Total Municipal Solid Waste (x10,OOO tons)

N W P wl QI 4 0 L-.

I I I I I I

9 \o 4 00 w 00