Full text of DOI:10.1080/10473289.2001.10464258

This article was downloaded by: [University Of Pittsburgh]
On: 13 October 2014, At: 11:38
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK
Journal of the Air & Waste Management Association
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/uawm20
Forecasting Generation of Urban Solid Waste in
Developing Countries—A Case Study in Mexico
Otoniel Buenrostro a , Gerardo Bocco b & Javier Vence c
a Instituto de Investigaciones sobre los Recursos Naturales, Universidad Michoacana de San
Nicolás de Hidalgo , Morelia , Mexico
b Departamento de Ecología de los Recursos Naturales , Instituto de Ecología, Universidad
Nacional Autónoma de México , Morelia , Mexico
c Centro Regional Universitario Centro Occidente, Universidad Autónoma de Chapingo ,
Morelia , Mexico
Published online: 27 Dec 2011.
To cite this article: Otoniel Buenrostro , Gerardo Bocco & Javier Vence (2001) Forecasting Generation of Urban Solid Waste
in Developing Countries—A Case Study in Mexico, Journal of the Air & Waste Management Association, 51:1, 86-93, DOI:
10.1080/10473289.2001.10464258
To link to this article: http://dx.doi.org/10.1080/10473289.2001.10464258
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions

Buenrostro, Bocco, and Vence
TECHNICAL PAPER
ISSN 1047-3289 J. Air & Waste Manage. Assoc. 51:86-93
Copyright 2001 Air & Waste Management Association
Forecasting Generation of Urban Solid Waste in Developing
Countries—A Case Study in Mexico
Otoniel Buenrostro
Instituto de Investigaciones sobre los Recursos Naturales, Universidad Michoacana de San Nicolás de
Hidalgo, Morelia, Mexico
Gerardo Bocco
Departamento de Ecología de los Recursos Naturales, Instituto de Ecología, Universidad Nacional
Autónoma de México, Morelia, Mexico
Javier Vence
Centro Regional Universitario Centro Occidente, Universidad Autónoma de Chapingo, Morelia, Mexico
ABSTRACT
INTRODUCTION
Based on a study of the composition of urban solid waste
(USW) and of socioeconomic variables in Morelia, Mexico,
generation rates were estimated. In addition, the genera-
tion of residential solid waste (RSW) and nonresidential
solid waste (NRSW) was forecasted by means of a mul-
tiple linear regression (MLR) analysis. For residential
sources, the independent variables analyzed were monthly
wages, persons per dwelling, age, and educational level of
the heads of the household. For nonresidential sources,
variables analyzed were number of employees, area of fa-
cilities, number of working days, and working hours per
day. The forecasted values for residential waste were simi-
lar to those observed. This approach may be applied to
areas in which available data are scarce, and in which there
is an urgent need for the planning of adequate manage-
ment of USW.
The amount and composition of the waste generated are
essential for adequate decision-making regarding the
management of urban solid waste (USW). Both charac-
teristics of waste vary with time and are affected by socio-
economic conditions. This aspect is most important for
planning management programs of USW in developing
countries, because the effect of socioeconomic variables
associated with the generation of solid waste is poorly
understood. Here, the effect of socioeconomic variables
such as income, level of consumption, and cultural and
educational environment on the generation of USW are
specific and vary from place to place.1,2
Most cities in developing countries undergo accel-
erated urbanization. One of the consequences of this
process is the transformation of the habits of consum-
ers. This change in consumer habits results in an in-
creased generation of waste, such as plastics and glass,
that is difficult to break down.3 The inadequate disposal
of this waste in the neighborhoods of the urban cen-
ters creates a strong, negative environmental impact on
soil, groundwater, the surrounding flora and fauna, and
public health.
IMPLICATIONS
Linear regression models based on socioeconomic vari-
ables are proposed to forecast USW generation. These
models can assist solid waste management planners with
efficient and easily applicable statistical instruments to
describe and forecast the generation of solid waste. This
is relevant in the development of solid waste manage-
ment plans. Accurate and detailed forecasts of solid waste
generation allow municipal authorities to plan capacity
requirements for waste-treatment systems and collection
and transportation systems, and to select sites and pre-
dict landfill lives. For policymakers and lawmakers, this
approach contributes to improving regulations about solid
waste generation.
In order to understand the USW generation, as well
as to propose management plans, it is necessary to con-
sider social and economic aspects. In developing coun-
tries, poverty of large sectors of the population is an
instrumental cause of the increase of social underdevel-
opment and marginality.4 The generation of USW is more
complex, and governments face difficulties in implement-
ing adequate waste management programs.
86 Journal of the Air & Waste Management Association
Volume 51 January 2001

Buenrostro, Bocco, and Vence
To understand, describe, and predict the unit waste
generation rate and the composition of USW, various
researchers have studied the influence of socioeco-
nomic factors. The variables more commonly analyzed
are community size; population density; monthly
wages per capita; number of persons per dwelling; per-
centage of urban population; annual average tempera-
ture; age, sex, and ethnicity of the population; size of
housing; geographical characteristics; land use; pro-
ductive activities; and communications.5-7 The over-
riding variable influencing unit waste generation rate
is population.2
Applicability of Forecasting Models in
Developing Countries: A Critical Review
The models and variables that have been shown to be
significant for developed countries have been extrapo-
lated to developing countries. In general, these models
do not consider the differences in behavior of the ex-
plaining variables involved, nor the scarcity of data about
the amount and composition of solid waste. Likewise,
in developing countries, they do not account for the
existence of informal practices, both in the generation
and in the collection of solid waste. Nevertheless, with
respect to developed countries, in developing countries
the level of recovery of components is higher.13 The re-
covery of waste in developing countries is mainly car-
ried out by marginal sectors of the population, by means
of a diversified range of informal activities, such as the
handpicking of materials during the collection of solid
waste, at the source of generation and at the dumping
grounds. This accounts for the difficulties involved in
precisely determining the volume of USW that is recov-
ered and recycled in these cases.
Solid Waste Generation Models
Solid waste generation models may be grouped into two
large classes:
(1) Descriptive generation models. These models pro-
vide information about the generation of solid
waste by different sources, such as residential,
commercial, institutional, and industrial sectors.
In general, these models are expressed in terms
of unit waste generation rates, which are factors
or multipliers relating the amount of waste gen-
erated with certain characteristics of the commu-
nity, such as total population size or number of
persons employed.7 Basically, generation data are
obtained by means of four different types of sam-
pling:8,9 sampling of waste generated by repre-
sentative sources (direct waste analysis), sampling
of the products derived from a given process of
waste management (waste product analysis), sam-
pling of the materials used for the production of
commodities by means of the enumeration of
supplies (market product analysis), and weight-
ing of representative samples of the loads of solid
waste collection vehicles (tonnage estimation
models).
(2) Predictive generation models. In general, these
models are based on unit waste generation
rates.7 The forecasting models are developed
using several statistical methods, such as the
correlation of the socioeconomic variables with
the generation of solid waste and the applica-
tion of linear regression models,5 principal
component analysis to suggest indicators of
potential generation of solid waste,10 the use
of nonparametric statistics to determine prob-
ability distribution for daily solid waste gen-
eration data,11 and the retrospective analysis by
time intervals to know the simultaneous influ-
ence of the variables involved in the genera-
tion of solid waste (geometric lag time series
analysis techniques).12
Despite the environmental problem posed by the in-
creasing generation of USW in developing countries, the
published research in this regard is scarce.14 Notwithstand-
ing the demographic growth, the expansion of industries,
the modification of food habits, and the process of ur-
banization in developing countries, the corresponding
authorities have not been urged to promote the develop-
ment of research leading to an understanding of the pat-
terns of generation of USW. This paper contributes to this
type of research; we present a case study aimed at fore-
casting the generation of USW in the city of Morelia
(Michoacan) in Mexico. As every major city in develop-
ing countries, Morelia is undergoing fast urban growth,
mostly driven by the immigration of rural population.
This is reflected in land use speculation, a chaotic expan-
sion of the urban areas at the expense of agricultural land,
and a sustained increase in the generation of USW. This
waste is becoming increasingly similar in its composition
to that generated in developed countries. However, the
management of USW remains to be dictated by obsolete
and inadequate criteria, something common to all devel-
oping countries. As a result, USW is disposed of in inad-
equate dumping grounds, following deficient planning
of USW management.15
The specific objectives of this research were (1) to
analyze the effects of monetary income, density of
dwellers per household, and educational level and age
in the generation of residential solid waste (RSW); (2)
to analyze the effects of the number of employees, the
size of facilities, and the number of daily working hours
and of working days in the generation of nonresiden-
tial solid waste (NRSW); and (3) to test the efficiency of
Volume 51 January 2001
Journal of the Air & Waste Management Association 87

Buenrostro, Bocco, and Vence
Table 1. Generation rates of urban solid waste (kg/capita/year).
these socioeconomic factors in the development of re-
gression models for the forecasting of USW generation
in fast-growing urban areas.
Source
kg/Capita/Year (Wet wt)
Residential
Commerce
Markets
230
219
2646
304
569
259
CONCEPTS
In this research, USW is defined as all solid waste gener-
ated within any human settlement. According to the
sources of generation, USW is divided into RSW, which is
produced by households, and NRSW, which is produced
by commerce, industries, and institutions/services. Each
one of these sources generate different waste, which is
defined accordingly, as follows:
Institutions/Services
Industry
Specials
(1) RSW: the solid waste generated in either single-
family or multifamily dwellings.
their own means, but dispose of them in the dumping
grounds used for all other USW.15
(2) CSW: the solid waste generated in commercial
facilities, department stores, supermarkets, res-
taurants, markets, and temporal ambulatory
markets (Mexican tianguis).
In most urban centers in these countries, with the
exception of the capital cities, large shopping malls are
few or absent. Additionally, these sources do not dispose
of their solid waste by means of the solid waste collection
services but, in general, they hire or own collection ser-
vices to transport their solid waste to dumping grounds.
Usually, this solid waste is highly appreciated by the mar-
ginal groups of the population because of its high con-
tent of metal and paper. The decomposing fruits and
vegetables in this solid waste are normally used to feed
domestic animals and, not infrequently, are also used for
human consumption.15
(3) Industrial solid waste: the solid waste generated
in all processes of extraction, transformation, and
production.
(4) Institutional and services solid waste: the solid
waste generated in private and governmental
offices, educational centers, museums, libraries,
archaeological zones, and recreation centers, such
as movie theaters and stadiums.
(5) Special solid waste: the solid waste requiring spe-
cial techniques for control, because of either the
relative hazard it represents or its particular con-
dition or state, or because it is required by the
standing legal regulations. This solid waste is gen-
erated in such sectors as research laboratories,
medical institutions or facilities, automotive or
industrial maintenance workshops, veterinary
facilities, drug stores, airports, and terrestrial
transportation terminals.16
METHODS
Sampling and Data Sources
During March and April 1998, a stratified sampling of 243
dwellings was randomly selected out of a total of 123,000
households. The number of samples per stratum was de-
termined by the Stein’s procedure.17 The sampling sites
were chosen by socioeconomic level following the classi-
fication of the Mexican bureau in charge of the census
(INEGI, by its Spanish acronym). This socioeconomic
stratification establishes three levels of monetary income:18
(1) low socioeconomic class: monthly income of less
than one minimum wage [about $90 (U.S. dol-
lars) per month in 1999];
The Urban Solid Waste in the Studied Area
In this city, the majority of the USW is generated by resi-
dential sources, thus corresponding to RSW. Table 1 pre-
sents the results of the sampling of waste in the studied
area. The contribution of nonresidential sources to the
generation of USW is due to the fact that, in general, the
micro and small industries (usually workshops) are pre-
dominant in the urban centers of developing countries.
Most of these sources mix their solid waste in the USW
stream because they use the municipal solid waste collec-
tion service. The medium and large industries (consider-
ing the number of employees, the size of their facilities,
and the volume of commodities production) are concen-
trated in industrial parks, which, in many cases, are sur-
rounded by marginal residential settlements. The medium
and large industries also transport their solid waste by
(2) middle socioeconomic class: monthly income
between one and two minimum wages [up to
about $180 (U.S. dollars) per month in 1999]; and
(3) upper socioeconomic class: monthly income be-
tween two and five minimum wages [up to about
$450 (U.S. dollars) per month in 1999].
In all cases, a questionnaire was simultaneously ap-
plied in order to obtain precise data about the monetary
income (income), the density of inhabitants per house-
hold (density), the age and educational level (education)
of inhabitants of the households, the land tenancy re-
gime of the household, and the availability of automo-
biles. The questionnaires were grouped by socioeconomic
88 Journal of the Air & Waste Management Association
Volume 51 January 2001

Buenrostro, Bocco, and Vence
level, and the answers were captured and analyzed in a
database (Excel version 1997).19 Further data were sub-
jected to a descriptive statistical analysis. Tables 2 and 3
show the results of the questionnaires for these sources.
Likewise, 32 nonresidential sources stratified by type
of economic activity were sampled. The total number of
samples included in the statistical analysis was determined
by the amount of samples retrieved from each selected
source. Nonetheless, the Stein’s procedure determined that
this number of samples was enough for the statistical analy-
sis. The samples of solid waste were taken during visits to
the facilities. In these cases, the questionnaires were aimed
at knowing the number of employees (employees), the area
of the facilities (area), the number of daily working hours
(hr), and the number of working days (days).
The socioeconomic data were considered as indepen-
dent variables, and data on generation of solid waste were
considered as dependent variables. Table 5 contains, for
residential and nonresidential sources, the names of the
analyzed variables together with their description. The
variables analyzed in this study were assessed simulta-
neously with the sampling of the solid waste generated
by each one of these selected sources. This was assumed
to be important to ensure a better relation between vari-
ables; it has been shown that if the solid waste is not as-
sessed simultaneously with the socioeconomic data, model
development may be affected.7
Predicting Models
The differences in socioeconomic conditions are reflected
in the composition of solid waste and in its generation
rate. Thus, an association between the generation of USW
and the corresponding socioeconomic variables is to be
expected.10 Based on this assumption, the Pearson product
moment correlation coefficient was initially applied to find
the coefficient of determination (R2) between the indepen-
dent variables. Subsequently, to test the hypothesis of in-
dependence between the selected explanatory variables, the
Student t test17 was applied on the R2 coefficient of the so-
cioeconomic variables. Because the analysis included more
than two variables, and these variables displayed a linear
distribution, a multiple linear regression (MLR) analysis was
For forecasting the generation of these sources, the
land tenancy regime of the household and availability of
automobiles were not included in the analysis because
these were highly correlated with income. Nonetheless,
it was observed that the availability of automobiles in the
low stratum might not correspond with income because
the majority of cars reported were old models. Also, it
was difficult to consider the age of all inhabitants of the
households; thus, it was decided to select the age of the
heads of the household only. Table 4 contains the num-
ber of nonresidential samples analyzed and the results of
the questionnaires applied to these sources.
Table 2. Socioeconomic data for residential sources.
Parameter
Socioeconomic Class
Middle
Low
High
Total number of households
61
125
57
Total number of persons
285
522
237
Average number of persons per household
4.7
4.2
4.1
Age distribution
Under 3
(%)
(%)
(%)
14
7
2
4 to 12
24
18
17
13 to 18
13
15
10
19 to 30
26
24
23
31 to 59
20
29
40
60 or over
2
7
8
Sex
(%)
(%)
(%)
Men
49
43
42
Women
51
57
58
Education level
Elementary school (not finished)
Elementary school
Junior high
High school
College
Men Women (%)
Men Women (%)
Men Women (%)
32
34
17
8
46
34
15
3
10
10
7
6
13
9
0
0
2
0
2
6
18
55
0
32
40
0
10
88
0
34
56
2
9
2
Graduate school
0
0
Volume 51 January 2001
Journal of the Air & Waste Management Association 89

Buenrostro, Bocco, and Vence
Table 3. Income data in residential sources.
(2) Mean square deviation regression (MSDR). The
lower the value, the better the fit of the model.
(3) The level of significance of the model (Pr > F).
For the model to be considered as significant, F
must have a value of less than α = 0.05.
Parameter
Socioeconomic Class
Low (%)
Middle (%)
High (%)
Incomea
$86 or less
$86–$91
$92–$264
$265–$609
$10–$965
$966 or more
House
Rent
33
12
53
2
0
0
0
RESULTS AND DISCUSSION
2
Forecasting Generation of Residential
18
28
27
25
4
Solid Waste
13
21
62
The estimated values of the R2 of the selected socioeco-
nomic variables were under 0.7, indicating poor
multicolinearity and independence among them.23 The
latter suggests that these variables are suitable for predict-
ing the generation of solid waste with this data set. How-
ever, the Student t test showed dependence between the
variables age versus income, educational level versus in-
come, and educational level versus age (Table 6). The above
facts had the effect of decreasing the significance of the
regression model when the four variables were included.
The analysis of variance (ANOVA) of the regression gave
an F value of 4.89, a Pr > F of 0.0008, an R2 of 0.075, and
an MSDR of 2.59. However, the variables age and educa-
tional level had nonsignificant regression coefficients,
with respect to their value of t (0.83 and 0.87, respec-
tively), indicating that these two variables did not explain
variability in per capita waste generation. This fact con-
firmed the results of the test for the hypothesis of inde-
pendence by means of the Student t distribution; likewise,
it corroborated the results of the questionnaires with re-
spect to the socioeconomic analysis. It was found that
the educational level increased with income, and this lat-
ter variable was inversely related to the number of dwell-
ers per household.15
0
0
10
61
26
3
14
58
24
4
17
79
4
Own
Paying
Lending
Car
0
Yes
7
75
25
93
7
No
93
aMonthly household income.
applied to determine the probable shape of the relation
between variables and to estimate the generation of solid
waste, which corresponds to the values of the analyzed so-
cioeconomic variables. From this, it may be ascertained that
the generation of USW may be explained by a multiple
linear equation having the form of eq 1.
Y = ßo + ß1X1 + ß2X2 + ß3X3 + ß4X4….. + ε
(1)
where Y is the dependent variable; ßo is the intercept; X1,
X2, X3, and X4,….are independent variables; ß1,ß2,ß3, and
ß4 are regression parameters; and ε is residuals.
The required statistical analyses were performed with
the Statistical Analysis System program (SAS).20-22 In order
to determine the forecasting model, which was most ap-
propriate for the generation of solid waste, the following
criteria were applied:
Because of the behavior described above, another
model was tested including the two variables whose values
of t were significant (income and density). This model did
not show an increase in the R2 coefficient, but the level of
significance showed a sensitive increase (0.0001), and a
decrease in MSDR was observed (2.57). These facts lead to
the decisive consideration of this model as a better fit than
the model with the four variables. The hypotheses stating
(1) Portion of sample variation explained (R2). An
ideal value of 1 for this coefficient indicates a
good fit of the model.
Table 4. Economic data in nonresidential sources.
Source
Samples
Number of
Employees
Size of
Facility (m2)
Schedule
(hr)
Working Days
(days/week)
per Source
Commerce
10
2
2.1
1.5
2
32.33
35.5
62
9.46
9.25
9.5
6.27
6
Institutions/Services
Microindustry
Special
4
6.25
6.06
6.17
16
2
28.7
36
8.6
Average
2
9.19
90 Journal of the Air & Waste Management Association
Volume 51 January 2001

Buenrostro, Bocco, and Vence
Table 5. Variables explaining the generation of USW in Morelia.
Variables Description
Independent Variables
Table 6. Values of t from the test for the independence hypotheses (Ho: ρ = 0 vs. Ha:
ρ ≠ 0) among explanatory variables (n–2 degrees of freedom, α = 0.05).
Compared Variables
Standard Value of t
Estimated Value of t
Residential Sources:
Residential Sources
X1
X2
X3
X4
Income: total monthly income per household (U.S. $).
Density: total number of dwellers per household.
Age: average age of heads of households (years).
Educational level: average number of years in an educational
institution attended by the heads of the households.
Density vs. income
Age vs. income
1.96
–0.627a
3.76b
1.96
Education vs. income
Age vs. density
1.96
10.17b
0.91a
1.96
Education vs. density
Education vs. age
1.96
–1.76a
2.39b
Nonresidential Sources:
1.96
X ’1
X ’2
Employees: the number of workers employed by the business.
Nonresidential Sources
Area: the total area in m2 occupied by the facilities
in which economic activity takes place.
Area vs. employees
Hr vs. employees
Days vs. employees
Hr vs. area
2.04
2.04
2.04
2.04
2.04
2.04
–1.26a
1.34a
0.144a
0.341a
–0.623a
0.845a
X ’3
X ’4
Hr: the total number of daily working hours.
Days: the total number of working weekdays.
Dependent Variables
Days vs. area
Y
RSW: the solid waste generated in households.
NRSW: the solid waste generated in the industrial,
commercial, institutional/services, and special sectors.
Days vs. hr
Y ’
aSignificant (Ho is not rejected ∴ independent); bNot significant (Ho is rejected ∴
not independent).
that the regression coefficients of the variables are equal to
zero (Ho: βI = 0 vs. Ha: βI ≠ 0) were rejected for estimators
βo, β1, and β2 at α = 0.05. This suggests that the proposed
model using the variables income and density may be ap-
plied to explain the MLR relation, and to forecast the gen-
eration of RSW, as shown in eq 2:
undoubtedly should affect the forecasting capability of
the model, although the socioeconomic variables selected
may be suitable. Yet, compared to the literature, these data
remain relevant.
Forecasting Generation of Nonresidential
Solid Waste
Y = βo+ β1X1 + β2X2 + ε
(2)
The test for the hypothesis of independence by means of
the Student t distribution with n–2 degrees of freedom
and α = 0.05 showed that all selected socioeconomic vari-
ables were independent (see Table 6). This suggests that
the optimal model for predicting the generation of solid
waste for the nonresidential sources is the one including
the four variables considered (employees, area, hr, and
days). However, the ANOVA of the MLR of the model with
all four variables was not significant (Pr > F = 0.17). Con-
sequently, runs were made using different models to de-
termine which was more adequate in terms of the level of
significance. From this analysis, it was determined that
the model including the variables X’3 (hr) and X’4 (days)
was optimal for the forecasting of NRSW. Inclusion of these
two variables increased the significance level to 0.05 and
the R2 to 0.177, and decreased the value of MSDR to 591;
thus, the model including the variables X’3 (hr) and X’4
(days) was adopted (eq 3).
Replacing values:
RSW = 1.02 + 0.000072X1 + 0.236X2
According to eq 2, the predicted value of RSW generated
is 2.46 kg, very close to the observed value of 2.49 kg.
Although the model is highly significant, the low value
of R2 (0.075) indicates that the variables suggested for fore-
casting RSW generation are yet of limited value in ex-
plaining the total variation in the data. However, in
previous studies using MLR to predict RSW components,
R2 coefficients between 0.26 and 0.57 have been reported.5
The models developed in this study provide a basis
to narrow the analysis of other factors involved in the
generation of USW, such as temperature. In a previous
study,24 a marked seasonal pattern in the generation of
USW in urban areas was reported. This was explained by
the predominance of some foodstuffs in certain seasons
of the year, and as drops and surges in consumption, which
derive from seasonal and economic fluctuations such as
the Christmas season. The data used in the present study
were gathered during a single season (spring), which
Y’ = β’o + β’3X’3 + β’4X’4
(3)
Replacing values:
NRSW = 0 + 0.19217X’3 + 0
Volume 51 January 2001
Journal of the Air & Waste Management Association 91

Buenrostro, Bocco, and Vence
The forecasted value of 1.77 kg is almost twice the
observed value of 0.925 kg; however, the model was highly
significant for forecasting the generation of NRSW. The
low value of R2 in the tested models indicates, as is the
case for RSW models, that the proposed variables for the
forecasting of generation are yet of limited value to ex-
plain the total variability of the data. One possible expla-
nation for this inadequacy of the forecasting models is
that the number of samples in the analyzed database,
despite being statistically significant, does not reflect the
total variation of the population studied. To improve the
forecasting capability of the models, the samples must be
expanded to a larger number of sources. It is important to
include additional socioeconomic variables, such as sales
volume. This variable is instrumental to assess the real
dimension of the economic activity of the sources. In-
deed, consideration of the size of the facility and the num-
ber of employees may lead to underestimating the degree
of economic activity of the source. However, data regard-
ing total sales volume are troublesome to obtain. In addi-
tion, sampling nonresidential sources is a difficult task
because of lack of cooperation from owners. The data-
base analyzed in this paper must be regarded as an initial
approach to the forecast of the generation of USW in ur-
ban areas that may be similar to that described here, and
of the variables that may be affecting their generation.
the explanation of these processes, they do offer an alter-
native with analytical value.
Our results indicate that, in Morelia, the generation
of residential waste differs significantly from that of non-
residential sources. In the case of the former, the vari-
ables which were found useful for forecasting the
generation of waste were monetary income and density
of dwellers per household; for the latter, the useful vari-
able was the number of daily working hours. The analysis
of these sources was difficult because of the restrictions
imposed by environmental legislation and policies. To
improve the forecasting power of the models, it is neces-
sary to expand the sampling to a larger number of sources.
It should be highlighted that the low number of samples
analyzed here followed the limited level of participation
of the sources. As a result, it is necessary to precede the
sampling with environmental education programs related
to the objectives of the analyses of generation of solid
waste. To this end, the different social sectors involved in
solid waste generation must work in coordination with
governmental agencies, chambers of commerce, services,
and industries. Emphasis must be on the confidentiality
and professionalism with which the volunteered infor-
mation will be utilized.
ACKNOWLEDGMENTS
Research on which the paper is based was granted by
Conacyt through a doctoral scholarship to the first au-
thor. The authors would like to acknowledge Dr. J.H.
Tanslconen of the Finnish Environmental Institute for
valuable comments to the paper, and the contribution of
Gonzalo Cortéz to the statistical analysis.
CONCLUSIONS
Adoption of adequate management measures to couple
with the environmental and public health impact caused
by the inadequate refusal of increasing urban waste is
urgent. The analyses of socioeconomic variables influ-
encing the generation of waste enable the forecast of
their quantity and are useful for planning their adequate
management. The models forecasting solid waste gen-
eration are useful analytic tools in the design of man-
agement programs. In the international range, a number
of models have been proposed that are based on the
analysis of several statistics and socioeconomic variables.
These models have shown to be efficient for the fore-
casting of solid waste generation in developed countries.
However, the applicability of these models is trouble-
some because of their theoretical complexity, as well as
the large data requirements. The use of linear regression
with efficient statistical instruments can explain and
forecast the generation of solid waste, based on data
obtained through relatively simple sampling designs, and
can do so at a low cost.
REFERENCES
1. Bowen, M.; Kontuly, T.; Hepner, G. Estimating Maquiladora Hazard-
ous Waste Generation on the U.S./Mexico Border; Environ. Aud. 1995,
19 (2), 281-296.
2. Hockett, D.; Lober, D.J.; Pilgrim, K. Determinants of Per Capita Mu-
nicipal Solid Waste Generation in the Southeastern United States; J.
Environ. Manage. 1995, 45, 205-217.
3. Richardson, G.M.; Whitney, J.B. Goats and Garbage in Khartoum,
Sudan: A Study of the Urban Ecology of Animal Keeping; Hum. Ecol.
1995, 23 (4), 455-475.
4. Leff, E. Ecología y Capital: Racionalidad Ambiental, Democracia
Participativa y Desarrollo Sustentable (Ecology and Capital: Ambient
Rationality, Participating Democracy and Sustainable Development);
Siglo XXI-UNAM (Universidad Nacional Autónoma de México):
Mexico City, 1994; p 437.
5. Ali Khan, M.Z.; Burney, F.A. Forecasting Solid Waste Composition—
An Important Consideration in Resource Recovery and Recycling;
Resource Conserv. 1989, 3, 1-17.
6. Matsuto, T.; Ham, R.K. Residential Solid Waste Generation and Recy-
cling in the USA and Japan; Waste Manage. Res. 1990, 8, 229-242.
7. McBean, E.A.; Fortin, M.H. A Forecast Model of Refuse Tonnage with
Recapture and Uncertainty Bounds; Waste Manage. Res. 1993, 11, 373-
385.
8. Bruner, P.H.; Ernst, W.R. Alternative Methods for the Analysis of
Municipal Solid Waste; Waste Manage. Res. 1986, 4, 147-160.
9. Rushbrook, P.E.; Finnecy, E.E. Planning for Future Waste Manage-
ment Operations in Developing Countries; Waste Manage. Res. 1988,
6 (1), 1-21.
Such tools are used in this research; their efficiency in
forecasting was assessed in a case study in Mexico. This
approach may be extrapolated to similar cases in other cit-
ies in developing countries. Although the models proposed
by this study are regarded as an initial approximation to
10. Cailas, D.M.; Kerzee, G.R.; Bing-Canar, J. An Indicator of Solid Waste
Generation Potential for Illinois Using Principal Components Analy-
sis and Geographic Information Systems; J. Air & Waste Manage. Assoc.
1996, 46, 414-421.
92 Journal of the Air & Waste Management Association
Volume 51 January 2001

Buenrostro, Bocco, and Vence
11. Arey, M.J.; Baetz, B.W.; Macdonald, P.D.M.; Byer, P.H. Use of Mixed
Probability Distributions for the Analysis of Solid Waste Generation
Data; Waste Manage. Res. 1993, 11, 387-402.
12. Ni-Bin, C.; Pan, Y.C.; Sen-Der, H. Time Series Forecasting of Solid
Waste Generation; J. Res. Manage. Tech. 1993, 21 (1), 1-10.
13. Heinen, J.T. A Review of, and Research Suggestions for, Solid Waste
Management Issues: The Predicted Role of Incentives in Promoting
Conservation Behavior; Environ. Conserv. 1995, 22 (2), 157-166.
14. Adedibu, A.A. A Comparative Analysis of Solid Waste Composition
and Generation in Two Cities of a Developing Nation; Environmen-
talist 1985, 5 (2), 123-127.
15. Buenrostro, O. Ph.D. Thesis, National Autonomous University of
Mexico, Mexico City, March 2000.
16. Residuos Sólidos-Terminología; Norma Oficial Mexicana NOM-AA-91-
1985; (Solid Waste Terminology; Mexican Official Standard NOM-
AA-91-1985); Secretaría de Comercio y Fomento Industrial (SECOFI):
Mexico City, Mexico, 1985; p 5.
17. Mendenhall, W.; Scheaffer, L.R.; Wackerly, D.D. Mathematical Statis-
tics with Applications; PWS Publishers: Belmont, CA, 1986; p 751.
18. XI Censo General de Población y Vivienda, 1990. Resultados Definitivos,
Tabulados Básicos (XI Census of Population and Housing 1990);
Instituto Nacional de Estadística Geografía e Informática (INEGI):
Aguascalientes, Mexico, 1991; p 2384.
19. Microsoft Excel 97; Microsoft Corp.: Redmond, WA, 1998.
20. SAS/IML Software: User’s Guide, release 6.03 ed.; Statistical Analysis
System Institute Inc.: Cary, NC, 1988.
21. SAS/IML Software: Usage and Reference, version 6, 1st ed.; Statistical
Analysis System Institute Inc.: Cary, NC, 1989.
22. SAS/GRAPH Software: Usage, version 6, 1st ed; Statistical Analysis Sys-
tem Institute Inc.: Cary, NC, 1991.
About the Authors
Otoniel Buenrostro (corresponding author) has a Ph.D. in
biology from the Universidad Nacional Autónoma de
México (UNAM). He is a junior researcher at the Instituto
de Investigaciones sobre los Recursos Naturales (INIRENA),
Universidad Michoacana de San Nicolas de Hidalgo. His
address is Apartado Postal 2-10, Morelia, Michoacán,
Mexico; phone: (4) 3272351; fax: (4) 3272350; e-mail:
otonielb@zeus.ccu.umich.mx. Gerardo Bocco is a senior
researcher at the Departamento de Ecología de los
Recursos Naturales (DERN), Instituto de Ecología, UNAM,
Campus Morelia, in charge of the Laboratory of
Geoecology. His address is Apartado Postal 27-3
(Xangari), 58089, Morelia, Michoacán, Mexico. Javier
Vence is an assistant professor in Statistics at the Centro
Regional Universitario Centro Occidente, Universidad
Autónoma de Chapingo. His address is Periférico
Independencia Poniente No. 1000, Colonia Lomas del
Valle, Morelia, Michoacán, Mexico.
23. Makridakis, S.; Wheelwright, S.C.; McGee, V.E. Forecasting: Methods
and Applications; Wiley: New York, 1983.
24. Buenrostro, D.O.; Bernache, P.G.; Cram, H.S.; Bocco, V.G. Análisis de
la Generación de los Residuos Sólidos en los Mercados de Morelia
(Solid Waste Generation Analysis in Markets of Morelia); Rev. Int.
Contam. Ambient. 1999, 15 (1), 27-32.
Volume 51 January 2001
Journal of the Air & Waste Management Association 93