You are on page 1of 14

HEALTH ECONOMICS

Health Econ. (2011)


Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/hec.1751

WEIGHT‐LOSS DIETING BEHAVIOR: AN ECONOMIC ANALYSIS

ODELIA ROSIN *
Department of Economics, COM Academic Studies, Rishon LeZion, Israel

SUMMARY
In light of the widespread phenomena of diet failure and excessive dieting, this paper presents a theoretical economic
analysis of the decision‐making process of weight‐loss dieting. The paper incorporates behavioral elements involved in the
process of dieting: effort exerted in dieting, influence of social norms concerning body weight, time‐inconsistent present
biased preferences, and a distinction between naiveté and sophistication. The model explains cyclic dieting and provides
interesting insights on the extent of weight‐loss dieting. The extent of dieting is an increasing function of initial body
weight and a decreasing function of the effort exerted in dieting and the strength of social norms concerning ideal weight.
Income and diet strictness have an ambiguous effect. In addition, greater dieting efforts are not necessarily balanced against
a slowdown in body metabolism or a higher initial body weight. Copyright © 2011 John Wiley & Sons, Ltd.

Received 16 August 2009; Revised 1 January 2011; Accepted 30 March 2011

KEY WORDS: Body weight; diet; obesity; social norm; weight loss

1. INTRODUCTION

At any given time, billions of adults and youth in the Western World attempt some form of weight‐loss dieting.
Considerable physical, mental and financial efforts are expended on dieting. The worldwide market for weight‐
loss products and services was 240 billion dollars in 2002.1 Nevertheless, dietary adherence rates are low and
long‐term success is limited.2 Weight loss is difficult to achieve; however, maintaining the achieved lower weight
presents an even greater challenge. Facing the trend of common dieting, we observed two phenomena, which
seemed to contradict each other. The first is that the world is getting fatter.3 The second phenomenon occurring at
the same time is that at least 10% of late adolescent and adult women report symptoms of eating disorders.4
In light of the widespread phenomena of diet failure and excessive dieting, this paper presents a theoretical
economic analysis of the decision‐making process of dieting. The model incorporates behavioral elements into
the modeling of dieting: personal effort exerted in dieting, the influence of social norms concerning body
weight, time‐inconsistent present biased preferences, and the idea of naiveté and sophistication.
There is a recent and growing economic literature trying to identify various biological, behavioral, and
environmental factors, which contribute to obesity.5 There is also a growing economic literature on policies and
interventions designed to curb obesity,6 yet few models on weight‐loss dieting appear in the economic literature:
a model of rational addiction (Dockner and Feichtinger, 1993), a dynamic model of weight cycles and unstable

*Correspondence to: Department of Economics, COM Academic Studies, 7 Yitzhak Rabin Ave., Rishon LeZion 75190, Israel.
E‐mail: rosino@colman.ac.il
1
Data source: Business Communications Co., Inc., RGA‐111R‐Weight Loss Market: Products, Services, Foods and Beverages, 2003.
2
As implied by nutrition literature (see, for example, Amigo and Fernandez, 2007; Bautista‐Castano et al., 2004; Byrne et al., 2003;
Dansinger et al., 2005; Elfhag and Rossner, 2005; Hoie and Bruusgaard, 1999; Popkess‐Vawter et al., 1998; Walsh and Flynn, 1995).
3
As reviewed in Rosin, 2008, section 2.2.
4
See, for example, www.nlm.nih.gov/medlineplus/eatingdisorders.html.
5
See, for example, Philipson and Posner, 1999; Cutler et al., 2003; Chou et al., 2004; Gruber and Frakes, 2006; Baum and Ruhm, 2009;
Courtemanche, 2009; Smith, 2009; for a broad survey on various explanations for obesity in the economic literature, see Rosin, 2008.
6
See, for example, Cawley et al., 2007; Nayga, 2008; Baum, 2009; Yaniv et al., 2009; Jason et al., 2010.

Copyright © 2011 John Wiley & Sons, Ltd.


O. ROSIN

overweightness (Levy, 2002), both solved using specific utility functions; models which analyze dieting
behavior by simulating food consumption choices of a representative woman (Suranovic et al., 2003; Suranovic
and Goldfarb, 2006); and a graphical model, which analyzes several causes of dieting (Goldfarb et al., 2006).
The present model explains weight cycling, which is a repeated loss and regain of body weight over time.
Other predictions of the model are that the extent of dieting is an increasing function of initial body weight and
a decreasing function of effort exerted in dieting, strength of social norms concerning ideal body weight,7 and
metabolism. Income and the strictness of the diet have an ambiguous effect. Another prediction is that greater
dieting efforts do not fully offset a metabolic slowdown or a higher initial body weight.
The paper is structured as follows: (i) the next section introduces the framework of the model; (ii) Section 3
provides its analytical solution including comparative statics; and (iii) Section 4 concludes.

2. THE FRAMEWORK

We deal with the problem of an individual who decides to undertake a diet to lose weight. The individual is
rational in the sense that he compares his costs and benefits and selects the best alternative to maximize his utility.
However, when he makes repeated decisions of the kind described in the model, he foresees only one period and
chooses his current actions without fully appreciating how his actions affect his future perceptions and behaviors.
This behavior can be explained by the theory of bounded rationality, in the sense of Akerlof (1991).
Alternatively, the behavior of the individual can be explained by time‐inconsistent present biased
preferences, in the sense of O’Donoghue and Rabin (1999). It is not only that people are impatient, which
means they want to receive rewards sooner and delay costs until later. When considering trade‐offs between
two future points in time, present biased preferences give stronger relative weight to the earlier point in time as
that point is approached.8
Two types of choices are identified within our decision‐making problem: the first involves immediate costs,
where the costs of an activity are immediate but any rewards are delayed; the second choice involves
immediate rewards, where the benefits of an activity are immediate, but any costs are delayed. The present bias
effect affects people in two opposite ways. When actions involve immediate costs, the present bias effect
causes people to procrastinate (i.e., wait when you should act). When actions involve immediate rewards, the
present bias effect causes people to act before the proper time.
Another distinction is between naïve and sophisticated people. Sophisticated people correctly predict their
future behavior. For example, they foresee that they will have self‐control problems in the future. Naïve people
do not foresee their self‐control problems. They always believe they will be time consistent in the future. They
plan to behave one way but, in fact, behave differently.
The previous distinctions are applied to the problem of weight control—eating is an immediate reward
activity as it involves immediate gratification, whereas the costs of gaining weight are delayed. On the other
hand, weight‐loss dieting is an immediate cost activity as it involves immediate costs, whereas the reward of
being slimmer is delayed. The belief conveyed in this model is that most people are naïve regarding weight‐
loss dieting as they are incorrectly optimistic about their future behavior.9
Consider an individual who obtains satisfaction from the consumption of food and other goods and has
disutility from being overweight. The disutility from being overweight is not something the individual is born
with. Rather, it is an insight that is gained during adolescence, early adulthood, or even later on in life. Until this
point, the individual is either not aware of the adverse effects of obesity on his physical appearance or health, or
these harmful effects are simply ignored. During the time leading up to this realization, the body weight of the

7
That is to say, social norms that favor being thin rather than being overweight.
8
Identical preferences were used by Laibson (1997), who used the term hyperbolic discounting.
9
There is a common saying ‘I will start my diet tomorrow’. In reality, we observe people continually delaying their diet yet another day.
However, while naïve people believe that when tomorrow comes they will really start dieting, sophisticated people know that they are
lying to themselves when they say they will start dieting tomorrow.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec
WEIGHT‐LOSS DIETING BEHAVIOR: AN ECONOMIC ANALYSIS

individual is the result of decisions made by his parents10 and his own decisions made prior to recognizing the
detrimental effects of being overweight. Thus, I assume that the initial weight of the individual prior to realizing
the disutility from being overweight and prior to undertaking any weight‐loss diets is exogenous.
However, even normal‐weight and lean adults who have been aware of the adverse effects of obesity on
their health, and have therefore maintained healthy eating habits, experience a slowdown in their metabolism
later on in life.11 When this occurs, they see that their body starts to accumulate fat. As a result, they now
realize they should make changes in their diet, either by choosing different kinds of food or by decreasing their
food intake.12 Therefore, even those people that have been aware of the adverse effects of obesity experience
an age‐related increase in body weight that triggers an increase in awareness and leads these people to diet for
the first time. Hence I assume that the initial weight prior to undertaking any weight‐loss diets later in life, like
the initial weight at the early stages of adulthood, is exogenous.
The question raised in this paper is neither why the individual became overweight nor why does he diet. Rather, it
is what influences the extent of dieting, given that he perceives himself as overweight and would like to lose weight.
The individual plans over a certain period (for example, a year from now) and would like to undertake a diet
at some point during the coming period. He decides what fraction of the period he would like to diet. The
duration of the diet is denoted by d, where 0 ≤ d ≤ 1.13 It is possible to decide to undertake a diet during the
whole period (i.e., d = 1). It is also possible to decide not to undertake a diet at all during the period (i.e., d = 0).
Food consumption is measured in calories. All kinds of food contribute calories (though different
quantities)14 to the total calorie consumption. This paper considers one type of weight‐loss diet: a low‐calorie
diet with a given amount of calories.15 Therefore, total calorie consumption throughout the period is
determined by the decision on d: the greater the duration of diet, the lower the overall caloric consumption. The
calorie consumption during the whole period would be minimal in the case of choosing d = 1 (that is, dieting
the whole period), which is the most restricting choice the individual can make concerning the duration of the
diet. This minimal quantity of calories is denoted by C d.
It is obvious that weight‐loss dieting involves considerable efforts: changing eating habits, setting up
boundaries on how much one eats, controlling the desire to eat high‐calorie foods, and so forth. The disutility
from exerting such efforts is described by:

E ¼ ð1 þ αðt ÞδÞe d; C d ; (1)
where t ≥ 1 counts the number of weight‐loss diets,16 α(t) > 017 expresses the additional effort needed in
repeated dieting, and δ is a dichotomous variable, which indicates whether the individual has dieted in the past:
δ = 1 if he has, or δ = 0 if he has not.
The assumptions on the disutility function, E, are as follows: (i) the disutility increases in the duration of the
diet and decreases in the calorie consumption of the diet because greater effort is exerted in order to comply either

10
Classen and Hokayem (2005) found strong evidence that a higher degree of obesity among mothers leads to a significantly increased
likelihood that they will raise an obese youth. Anderson and Butcher (2006) explain that parents may pass along to their children a
susceptibility to being overweight. Parents influence their child’s weight in two ways: genetics and behavior. Children’s eating habits are
affected by the food selection in their household (for example, energy‐dense foods versus fruits and vegetables). In addition, the
children’s physical activity can be affected by how active their parents are.
11
As people age, metabolism typically begins to slow down because of a loss of muscle mass and an increase in body fat (see for example,
Mattson, 2010; Nedergaard and Cannon, 2010; Pellizzon et al., 2000). Moreover, menopause in women is associated with an increase in
total fat and accelerated accumulation of abdominal body fat that is independent of total adiposity and exceeds changes normally
attributed to aging (see for example Morita et al., 2006; Toth et al., 2000; Tchernof and Poehlman, 1998).
12
Goldfarb et al. (2006) called such a motive for dieting ‘aging‐associated dieting’ and explained that it involves shifts in weight as the
basal metabolic rate decreases with age.
13
For example, a 6‐month diet out of one year would obtain d = 0.5.
14
According to the caloric density of each kind of food.
15
It is unnecessary to consider other weight loss diets as no significant difference was found. In the long run, among low carbohydrate
diets, low‐fat diets, low‐glycemic load diets, and low‐calorie diets (see, for example, Dansinger et al., 2005).
16
t = 1,2,3,…
17
Strictly positive for every t.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec
O. ROSIN

with a longer diet or with a low‐calorie diet, that is, ed > 0 and eCd < 0; (ii) the disutility increases at decreasing
rates as calorie consumption of the diet rises, that is, eCd Cd < 0; (iii) the disutility increases at increasing rates as
the duration of the diet rises, that is, edd > 0; (iv) the joint derivative is negative, meaning that when calorie
consumption during the diet is lower, an additional day of dieting boosts the effort, that is, edCd < 0; and (v)
greater effort is needed from an individual with a ‘richer’ history of repeated dieting, that is, Δα Δt > 0.
When dieting decisions are repeated,18 each diet actually causes a change in this disutility.19 When our
individual takes his decision at the first period he does not foresee the change in his utility function. Hence he
does not foresee that the next diet will need more effort than the current one. Specifically,20 a sophisticated
individual recognizes that the previous assumption (v) is true, whereas a naïve individual falsely believes that
Δα
Δt ¼ 0. That is, the naïve individual does not believe that the next diet will need more effort than the current one.
Suppose the individual had not cared about his weight at all, thus he neither had any utility nor any disutility
from being overweight.21 In this case, he would obtain utility exclusively from consumption of food and other
goods. He would then maximize utility subject to his budget constraint and consume an optimal quantity of
calories denoted by CN. Hence CN is the maximal quantity of calories that an individual who does not care about
his weight would consume, given his budget constraint.22 Essentially C N is positively correlated with income.23
Now, suppose for the rest of the model, that our individual does consider his weight and thus has disutility
from being overweight. His calorie consumption for the dieting fraction of the period equals dC d, and for the
non‐dieting fraction of the period, his calorie consumption equals (1 − d )C N. Therefore, the total consumption
(throughout the whole period the individual foresees) can be expressed as a function of the diet duration:
C ðd Þ ¼ dC d þ ð1− d ÞC N ; (2)
and is actually a weighted average of C and C . The consumption function satisfies Cd = C − C < 0
d N 24 d N

and Cdd = 0.
Accordingly, the diet duration d can be expressed as a function of the total consumption:
CN − C
d ðC Þ ¼ : (3)
CN − Cd
Naturally, changes in body weight result not only from calorie consumption but also from calorie
expenditure. Calories are expended in physical activity but also when the body is at rest through basal
metabolism.25 Suppose that a certain level of calorie consumption, denoted by Ĉ, balances against calorie
expenditure from physical activity and basal metabolism.26 Thus, consuming Ĉ maintains current body weight,
whereas consuming more than Ĉ results in weight gain.27
For the duration of the diet, the individual will certainly lose weight. However the duration of the diet is
merely a fraction of the whole period. The whole period, which consists of the dieting fraction and the

18
Goldfarb et al. (2006) gave an example of yearly cyclical dieting every spring in order to be able to fit into bathing suits, and called such
a pattern of dieting ‘Style‐provoked dieting’. Another example for repeated decisions is choosing d = 1 over two consecutive periods of a
year, implying a continual diet that lasts 2 years. In both examples, one dieting decision is taken every period/year.
19
Akerlof, 1991 has discussed the view that people do not have fixed utilities but rather have utilities that do change. People commonly fail
to fully foresee those changes.
20
Following O’Donoghue and Rabin, 1999.
21
In fact, this was the typical situation a hundred years ago when food was scarce and people ate what they could afford.
22
Assuming that C N > C d, otherwise the individual would not become overweight in the first place.
23
See Appendix A for the formulation.
24
For example, if the individual decides to undertake a diet during half of the period, his total calorie consumption during the period would
be 0.5Cd + 0.5CN. In general, Cd ≤ C ≤ CN for 0 ≤ d ≤ 1.
25
Basal metabolism relates to vital physiological functions that keep the body alive while fasting and at total rest (respiration, circulation,
digestion, cell renewal, body temperature regulation, and so forth). The energy required for basal metabolism is measured by basal
metabolic rate (BMR), see Harris and Benedict, 1918. See also Cutler et al., 2003.
26
Assuming that the levels of basal metabolism and physical activity are stable throughout the given period the individual foresees.
27
Notice that consuming CN implies gaining weight if C N > Ĉ.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec
WEIGHT‐LOSS DIETING BEHAVIOR: AN ECONOMIC ANALYSIS

complementary non‐dieting fraction, will not necessarily end in weight loss. Moreover, it may even end in
weight gain; in case total calorie consumption exceeds total calorie expenditure during the whole period.28
Body weight can possibly be sustained throughout the whole period. There exists a certain  duration of diet,
which implies a total calorie consumption that maintains initial weight, that is, d̂ ¼ d Ĉ . It is derived by
setting Ĉ in equation 3:
C N − Ĉ
d̂ ¼ : (4)
C N −C d
In order to achieve weight loss at the end of the period, the individual has to consume less than Ĉ calories
throughout the whole period, thus choosing d > d̂. Otherwise, the total consumption exceeds Ĉ, and the extra
calories will overall result in weight gain despite the diet.
It is assumed that consuming one extra calorie ( beyond Ĉ ) results in a weight gain of β.29 Hence the body
weight at the end of the period is

W ¼ W0 þ β C− Ĉ ; (5)
where W0 is the initial body weight (at the beginning of the period).30 Notice that body weight cannot, at any
stage, fall below a minimum body weight essential for living, denoted as W _ > 0.
In order to determine how much weight one should lose, it is possible to compare the actual weight with the
standard absolute weight measures.31 Nevertheless, one usually takes his dieting decisions according to how
overweight he perceives himself to be.32 The subjective way, in which he perceives himself, is influenced by
social norms and expectations regarding weight.33 Hence it is assumed that the individual compares his actual
weight, W, to a subjective ideal weight, W I, which he thinks he should weigh. The insight on this ideal weight is
developed along with realizing that being overweight has adverse effects on his physical appearance or health.34
The overweightness of the individual is defined as the gap between his actual weight and his subjective
ideal weight:
W g ¼ W−W I (6)
where W I is assumed to be constant throughout the model.
The individual has disutility from deviating from his subjective ideal weight.35 As this paper deals with
dieting decisions, its focus is on those who view themselves as overweight, that is, W g > 0.36 The disutility
from the weight gap is described by Y(W g ), where (i) the disutility increases in the weight gap; that is, Y′ > 0;
and (ii) the disutility increases at increasing rates as the weight gap rises, that is, Y″ > 0.

28
For example, even though one could lose a few pounds during 1 month of the year, he could gain more than he lost for the rest of the
year.
29
Cutler et al. (2003) reported that for a typical person, an increase in calorie intake of 3500 cal increases weight by 1 lb.
30
Goldfarb et al. (2006) regard weight loss as the difference between two optimal states of weight. However, they ignore characteristics of
the dieting process per se.
31
The international guidelines on body mass index (BMI) offer cutoffs of 25 and 30, which define overweight and obesity, respectively (see
Rosin, 2008, section 2.1).
32
For the importance of subjective weight perceptions, see Brener et al., 2004; Neumark‐Sztainer et al., 2002; Oswald and Powdthavee,
2007a, b.
33
Socio – cultural norms of appearance are considered in Burke and Heiland, 2006; Eisenberg et al., 2005; and Levy, 2002. Goldfarb et al.,
2006, discuss desired weight yet focus on the motives for dieting.
34
In other words, realizing there is a disutility from being overweight.
35
Oswald and Powdthavee (2007a, b) pointed out that half the British population view themselves as overweight, and suggested a theory of
obesity imitation where utility depends on relative weight. The general concept that well‐being depends on relative measures appears in
the economic literature (see, for example, Easterlin, 2001).
36
This implies that losing weight increases utility. However, Philipson and Posner (1999) assumed that the utility function is non‐
monotonic in weight, having an inverted U‐shape around an ideal weight the individual would prefer to have. Deviating from the ideal
weight by either gaining or losing weight reduces utility. Our overweight individual may be perceived as being located on the rightward
wing of the inverted U‐shape utility curve, where utility decreases monotonically with higher weight.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec
O. ROSIN

The utility of the individual is defined as:


U ¼ U ðC; Z; W g ; d Þ ¼ V ðC; Z Þ−Y ðW g Þ−E ðd Þ; (7)
where Z is an aggregate of all non‐food consumption.
The utility function is partially additive,37 continuous, twice differentiable, and satisfies VC > 0, VCC < 0,
VZ > 0, VZZ < 0 and VCZ > 0.
The budget constraint is given by: 38
CþZ ¼I (8)
where I is income.

3. THE SOLUTION

The objective of the individual is to maximize his utility with respect to the diet duration, given his budget
constraint. The duration of the diet, d, has opposite effects on the utility of the individual. On the one hand, a
longer diet decreases utility by obtaining less satisfaction from eating and by exerting more effort. On the other
hand, a longer diet increases utility by releasing budget resources for other consumption39 and by reducing
disutility from overweightness.
Because the decision on the diet duration (d) actually determines the total amount of calories consumed
during the whole period (C), maximizing utility with respect to C is equivalent to maximizing it with respect to
d. Substituting (1), (3), (5), (6), and the budget constraint (8) into (7), utility is rewritten as a function of C:
 
U ðC Þ ¼ V ðC; I−C Þ−Y W0 þ β C− Ĉ −W I −ð1 þ αðt ÞδÞeðdðCÞÞ: (9)
The first order condition to the above maximization problem is:
1 þ αðt Þδ
UC ¼ VC −VZ −βY ′ þ ed ¼ 0: (10)
C N −C d
Let C* be the solution to (10), that is, C* is the optimal amount of calories to be consumed during the whole
N
−C*
period, hence d* ¼ CCN −C d is the optimal duration of dieting.

The second order derivative is:


1 þ αðt Þδ
UCC ¼ VCC −2VCZ þ VZZ −β 2 Y ″ − edd < 0: (11)
ðC N −Cd Þ2
Based on the model’s assumptions, the second order condition for maximization is satisfied.
The next sections determine, by comparative static analyzes, how the optimal duration of dieting, d*, and
the total consumption, C*, are affected by changes in the parameters of the model: initial body weight, effort
needed in repeated dieting, body metabolism, income, strictness of the diet, and subjective ideal weight.
∂d 40
Technically, the comparative statics are carried out for C*. Then, the result is multiplied by ∂C to obtain
*
comparative static results for d . Wherever the effect on body weight at the end of the period is not clear, it is
examined as well.41

37
Assumed for tractability.
38
Assuming that the price of C is normalized to 1; thus Z actually denotes the expenditure on non‐food consumption.
39
In reality, weightloss diets do not necessarily release budget for other consumption because some calorie‐dense foods, which are
relatively cheap may be substituted for low‐calorie foods like fruits and vegetables, which are relatively expensive. However the model
was not designed to capture the pricing of different types of foods, as this is not the focus of the paper.
∂d
40
With the exception of changes in CNor Cd, differentiating Equation (3) yields ∂C ¼ CN−1
−C d
.
41
Specifically for initial weight and body metabolism. Regarding other parameters, it is obvious that an increase in total calories implies a
rising weight.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec
WEIGHT‐LOSS DIETING BEHAVIOR: AN ECONOMIC ANALYSIS

1
A

d̂ S
F
0 W0

Figure 1. Repeated decisions according to the belief of the naïve individual: dieting more than d̂ implies losing weight, whereas dieting
less than d̂ implies gaining weight. Therefore point S is the steady state at which the extent of dieting d̂ maintains a stable weight

3.1. Initial weight


Consider a change in the initial weight of the individual, W0. Equation (10) is totally differentiated (see
∂C ∂d
Appendix B, Equations A6 and A7) to obtain ∂W 0
< 0 and ∂W 0
> 0. Hence the higher the initial weight of the
individual, the longer the duration of dieting and the lower the calorie consumption.42
Regarding the effect on body weight at the end of the period, we might expect the longer dieting and the lower
calorie consumption to offset the higher initial weight and result in a lower weight. However equation (A8)
∂W
in Appendix B yields ∂W 0
> 0; hence the higher the initial weight of the individual, the higher his weight at the end
of the period, despite the longer dieting and his greater effort.
Let us examine a situation of repeated decisions over a few periods, and assume for a moment that Δα Δt ¼ 0,
according to the belief of the naïve individual. The d ¼ d̂ line (see Figure 1) represents the diet duration, which
maintains a stable weight (see Equation 4). Suppose the individual starts above the line d ¼ d̂ (for example, at
point A), acts according to his belief and takes repeated decisions. Being above the line means dieting more
than d̂ , thus losing weight during the period. Over time the individual will lose more and more weight until he
converges to point S (which lies on the d ¼ d̂ line). Now suppose the individual starts below the d ¼ d̂ line (for
example, at point F). Being below the line means dieting less than d̂ and thus gaining weight during the period.
Over time the individual will gain more and more weight until he converges to point S. At point S, he will
choose d̂ and thus maintain a stable weight. Therefore S is a steady state. Actually at point S the weight does
not equal the ideal weight; hence S reflects a steady state of overweightness.

3.2. Effort exerted in repeated dieting


It is interesting to know whether the duration of the diet is influenced by the effort one has to exert in order to
engage in repeated dieting. Consider a change in α, the effort exerted in repeated dieting. Equation (10) is
totally differentiated (see Appendix B, Equations A9 and A10) to obtain ∂C ∂d
∂α > 0 and ∂α < 0.
As it was initially assumed that additional effort is needed in repeated dieting, i.e., Δα
Δt > 0, each additional
diet causes an increase in α. Hence as the effort rises, the individual will choose to reduce dieting. For example,
if he has exerted effort in a long diet last year, he may perceive himself as unable to comply with another long
diet and would thus choose a shorter diet.
Now let us take into consideration the above analysis of initial weight together with the increasing effort in
repeated dieting, according to the belief of the sophisticated individual (which is similar to the actual situation
of the naïve individual, contrary to his belief). Suppose that at the first period, the individual is on the p1 curve

42
This prediction disagrees with the common notion that obese people do not make an attempt to lose weight. Our prediction is consistent
with the empirical findings of Ikeda et al., 2004, saying that women with higher BMIs tend to start dieting before age 14, have dieted
more frequently than women with lower BMIs, and many of them have a history of dieting failure.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec
O. ROSIN

p1 p2
1
A
B
R
d̂ S
B'

0 W0
W0A

Figure 2. Repeated decisions according to the belief of the sophisticated individual: at the second period, a greater effort is needed. Hence
instead of moving from point A to B and later converging to S, the individual will move to point B′ on a new curve p2, then converge to a
new steady state, R, which implies a higher weight

(see Figure 2) above the d ¼ d̂ line, and takes repeated decisions. He chooses to undertake a diet for more than
d̂, and thus loses weight. Suppose, for example, an individual whose weight is W0A (see Figure 2, point A). At
the second period, a greater effort is needed,43 causing a greater disutility. Hence the individual will move to a
new curve, denoted by p2, which is below (to the right of ) the curve of the first period. Instead of moving to
point B on the curve p1, he will move to point B′ on the new curve p2, and find himself below the d ¼ d̂ line.
Then, at the second period he will diet less than d̂, thus regain weight. After a few periods, instead of
converging to S, he will converge to a new steady state R, which reflects a higher weight, higher than at point S
and even higher than at the initial point A. This process demonstrates weight cycling, which is the repeated loss
and regain of body weight.

3.3. Metabolism
Do changes in body metabolism affect the extent of dieting? Personal body metabolism is represented by two
parameters, Ĉ and β . Ĉ is the consumption level that maintains current body weight, whereas β is the weight
gain from consuming one extra calorie.
First, consider a change in the amount of calories the individual needs to consume in order to sustain his
current weight. When Ĉ decreases, the individual has to eat less to maintain his current weight, this implies a
metabolic slowdown.44 Equation (10) is totally differentiated (see Appendix B, Equations A11 and A12) to
obtain ∂∂CĈ > 0 and ∂∂dĈ < 0. Hence the more the metabolism slows down, the longer the diet duration.
Regarding the effect on body weight at the end of the period, we might expect the longer dieting and the
lower calorie consumption to offset the metabolic slowdown and result in a lower body weight. However
Equation (A13) in Appendix B yields ∂W ∂ Ĉ
> 0; hence the more the metabolism slows down, the higher the
weight at the end of the period, despite the longer dieting and the greater effort.
Now consider a change in the weight gain caused by consuming one extra calorie.45 When β increases, the
individual gains more weight from eating extra calories, and this is another sense of a metabolic slowdown.
Equation (10) is totally differentiated (see Appendix B, Equations A14 and A15) to obtain ∂C ∂d
∂β < 0 and ∂β > 0.
Hence the higher the weight gain caused by consuming one extra calorie, the longer the diet duration.
As for the impact on body weight at the end of the period, equation (A13) in Appendix B yields an ambiguous
sign for ∂W ∂β ; hence in this case we do not know whether the longer dieting dominates the metabolic slowdown.

43
The additional effort stemming from the term α(t)δ in Equation (1).
44
A slowdown in metabolism is a natural consequence of aging. See footnote 11.
45
Beyond Ĉ.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec
WEIGHT‐LOSS DIETING BEHAVIOR: AN ECONOMIC ANALYSIS

3.4. Income
Do changes in income influence the extent of dieting? Consider a change in the individual’s income, I. Notice
that the positive correlation of CN with income has to be taken into account.46 An increase in CN affects the
optimal calorie consumption in opposite directions.
Totally differentiating Equation (10) yields that both ∂C ∂d
∂I and ∂I are ambiguous (see Appendix B, Equations
A17 to A18). Hence the influence of income changes on the extent of dieting is ambiguous. Total calorie
consumption and the extent of dieting can either be normal or inferior goods for our individual.47
The ambiguous component of (A17) relates to the marginal effort. It is the only component that might
modify the effect of income changes on the total amount of calories from positive to negative.

3.5. Diet strictness


Weight‐loss diets are actually not all the same. As some are stricter than others, we now ask whether diet
strictness influences the extent of dieting. Consider a change in Cd, the quantity of calories consumed during
dieting. This quantity can be used as a reverse proxy for the strictness of the diet. A decrease in Cd implies a
stricter diet, which is harder to comply with. A stricter diet requires more effort on the one hand, but results in a
greater weight loss on the other.
Totally differentiating Equation (10) (see Appendix B, Equations A19 to A20), an ambiguous sign is obtained for
∂C ∂d
both ∂C d and ∂C d . Hence we do not know the influence of a stricter diet on the total amount of calories and on the

extent of dieting. A stricter diet increases marginal effort, thus decreases the total amount of calories. However a
stricter diet could achieve a desired weight loss in a shorter time, decrease effort, and raise the total amount of calories.

3.6. Social norms concerning weight


A very interesting question is whether the choice of dieting is influenced by social norms. Modern western
societies put an emphasis on body shape and weight. We all notice the cultural values of thinness, as
represented by extremely underweight fashion models. How do these perceptions of beauty influence the
choice of dieting? Consider a change in the subjective ideal weight, W I, which reflects social norms
concerning weight. Equation (10) is totally differentiated (see Appendix B, Equations A21 and A22) to obtain
∂C ∂d
∂W I > 0 and ∂W I < 0. Hence the more the ideal weight decreases the greater the increase in diet duration.
This analysis clearly demonstrates the influence of social norms concerning weight on the choice of dieting.
The more a person is pressured to imitate the mainstream image the greater the desire to be ‘thin and perfect’. The
gap between reality and the ideal of beauty leads to pressures to be thin and to an increased risk for developing
body image dissatisfaction and making extreme dieting efforts.48 Frequent attempts of dieting, despite the low
probability to succeed in losing weight, may result in developing eating disorders in the long run. The model
predicts that an increase in ideal weight would cause the individual to choose to reduce dieting. Therefore,
actions for developing more realistic weight norms might moderate the unhealthy extreme dieting efforts.

4. CONCLUSION
The present paper has focused on the weight‐loss dieting process, given that the individual perceives himself as
overweight and would like to undertake a diet to lose weight. As behavioral aspects play an important role in
this process, incorporating behavioral features enables a better understanding of the essence of dieting.

46
See Appendix A.
47
In regard to income and weight, Cawley et al. (2010) found no significant effect of income on weight. This recent finding is opposed to
previous empirical findings on higher rates of obesity among low income people (See Rosin, 2008, section 2.12).
48
Extreme weight control behaviors (such as vomiting and taking diet pills, laxatives, or diuretics) were found among overweight youth,
particularly adolescent girls. Such behaviors predict health outcomes related to eating disorders (see, for example, Neumark‐Sztainer
et al., 2002, 2006).

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec
O. ROSIN

The model derives interesting insights on the extent of weight‐loss dieting. Higher initial body weight and
also stronger social norms of low body weight increase the duration of dieting, whereas more dieting
experience decreases the duration of dieting. Income and diet strictness have a more complex effect. Greater
dieting efforts do not fully offset the effects of a metabolic slowdown or a higher initial body weight; hence
body weight at the end of the period is higher despite the longer dieting and the greater efforts.
The basic modeling provided in this paper could be extended in various directions. Hopefully insights
derived from this essay may contribute to developing public policies for inducing healthy weight‐loss efforts
and implanting some of the common extreme and unhealthy dieting behaviors.

APPENDIX A

Consider an individual who obtains utility from consumption of food and other goods (without taking weight
and dieting into consideration). His objective is to maximize
V ðC; Z Þ; (A1)
where C is food consumption, and Z is an aggregate good of all non‐food consumption, under the budget
constraint
C þ Z ¼ I; (A2)
where I is income.49
The utility function is continuous, twice differentiable, and satisfies VC > 0, VCC < 0, VZ > 0, VZZ < 0, and
VCZ > 0.
The first order condition is
VC −VZ ¼ 0: (A3)
Let CN be the solution to the previous maximization problem.
The second order derivative is
VCC −2VCZ þ VZZ < 0 (A4)
Based on the model’s assumptions, the second order condition for maximization is satisfied. It is now
possible to show that CN is positively correlated with I:
∂CN VCZ −VZZ
¼− >0 (A5)
∂I VCC −2VCZ þ VZZ

APPENDIX B
COMPARATIVE STATICS
Totally differentiating Equation (10) with respect to the initial weight of the individual obtains
∂C βY ″
¼ < 0; (A6)
∂W0 UCC
based on Equation (11) and the model’s assumptions. Thus:
∂d 1 ∂C
¼− N d⋅ > 0: (A7)
∂W0 C −C ∂W0

49
Assuming that the prices of C is normalized to 1; thus, Z actually denotes the expenditure on non‐food consumption.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec
WEIGHT‐LOSS DIETING BEHAVIOR: AN ECONOMIC ANALYSIS

As we can show that A6 is smaller than 1, we also obtain


∂W ∂C
¼1þβ > 0: (A8)
∂W0 ∂W0
Totally differentiating Equation (10), with respect to the effort exerted in repeated dieting, obtains
∂C δed
¼− > 0; (A9)
∂α UCC ðC N −Cd Þ
based on Equation (11) and the model’s assumptions. Thus,
∂d 1 ∂C
¼− N d⋅ < 0: (A10)
∂α C −C ∂α
Naturally, this change is relevant only in case the individual has dieted in the past, that is, δ = 1, otherwise
both derivatives are equal to 0.
Totally differentiating Equation (10), with respect to the amount of calories needed to maintain current body
weight, obtains
∂C β2Y ″
¼− > 0; (A11)
∂ Ĉ UCC
based on Equation (11) and the model’s assumptions. Thus,
∂d 1 ∂C
¼− N d⋅ < 0: (A12)
∂ Ĉ C −C ∂Ĉ
As we can show that A11 is smaller than 1, we also obtain
 
∂W ∂C
¼β −1 < 0: (A13)
∂ Ĉ ∂Ĉ
Totally differentiating Equation (10) with respect to the weight gain from consuming one extra calorie, obtains:
∂C Y ′ þ βCY ″
¼ < 0; (A14)
∂β UCC
based on Equation (11) and the model’s assumptions. Thus,
∂d 1 ∂C
¼− N d⋅ > 0: (A15)
∂β C −C ∂β
We also obtain that
 
∂W ∂C β
¼C 1þ ⋅ (A16)
∂β ∂β C
is ambiguous.
Totally differentiating Equation (10) with respect to the income obtains
 N 
e ∂C ðC−C d Þ
N
e ∂C
−VCZ þ VZZ −ð1 þ αðt ÞδÞ ddðC∂IN −Cd Þ3 − ðCNd−C∂I d Þ2
∂C
¼ ; (A17)
∂I UCC
 N  
∂C ∂C
C N −C d − ∂C∂I ðC N −C Þ
N
∂d ∂I − ∂I
¼ : (A18)
∂I ðC N −C d Þ2
Based on Equation (11) and the model’s assumptions, both derivatives are ambiguous.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec
O. ROSIN

Totally differentiating Equation (10) with respect to the diet strictness obtains:
1þαðtÞδ 
∂C ed
C N −C d C N −C d
þ edCd
¼− ; (A19)
∂C d UCC
∂C

∂d − ∂C d C −C
N d
þ C N −C
¼ : (A20)
∂C d ðC N −C d Þ2
Based on Equation (11) and the model’s assumptions, both derivatives are ambiguous.
Totally differentiating Equation (10) with respect to the ideal weight obtains
∂C βY ″
¼ − > 0; (A21)
∂W I UCC
∂d 1 ∂C
¼− N d⋅ < 0: (A22)
∂W I C −C ∂W I
based on Equation (11) and the model’s assumptions.

ACKNOWLEDGEMENTS

This study has been financially supported by Bar‐Ilan University, the President Scholarships. This paper is
based on a part of my PhD dissertation at Bar‐Ilan University (approved 2008). I am grateful for the
constructive assistance of Arye L. Hillman and Nava Kahana from the Economics department. I also thank Anat
Alexandron‐Lavon, Gideon Yaniv and two anonymous referees for their valuable comments and suggestions.

REFERENCES

Akerlof GA. 1991. Procrastination and obedience. The American Economic Review 81(2): 1–19.
Amigo I, Fernandez C. 2007. Effects of diets and their role in weight control. Psychology, Health & Medicine 12(3):
321–327.
Anderson PM, Butcher KF. 2006. Childhood obesity: trends and potential causes. The Future of Children 16(1): 19– 45.
Baum CL. 2009. The effects of cigarette costs on BMI and obesity. Health Economics 18: 3–19.
Baum CL, Ruhm, CJ. 2009. Age, socioeconomic status and obesity growth. Journal of Health Economics 28(3): 635– 648.
Bautista‐Castano I, Montoya‐Alonso JA, Serra‐Majem L. 2004. Variables predictive of adherence to diet and physical
activity recommendations in the treatment of obesity and overweight, in a group of Spanish subjects. International
Journal of Obesity and Related Metabolic Disorders 28(5): 697–705.
Brener ND, Eaton DK, Lowry R, McManus T. 2004. The association between weight perception and BMI among high
school students. Obesity Research 12(11): 1866–1874.
Burke M, Heiland F. 2006. Social Dynamics of Obesity. Federal Reserve Bank of Boston, Public Policy Discussion Paper
No. 06 –5.
Byrne S, Cooper Z, Fairburn C. 2003. Weight maintenance and relapse in obesity: a qualitative study. International
Journal of Obesity and Related Metabolic Disorders 27(8): 955–962.
Cawley J, Meyerhoefer C, Newhouse D. 2007. The impact of state physical education requirements on youth physical
activity and overweight. Health Economics 16(12): 1287–1301.
Cawley J, Moran J, Simon K. 2010. The impact of income on the weight of elderly Americans. Health Economics 19:
979–993.
Chou SY, Grossman M, Saffer H. 2004. An economic analysis of adult obesity: results from the behavioral risk factor
surveillance system. Journal of Health Economics 23(3): 565 –587.
Classen T, Hokayem C. 2005. Childhood influences on youth obesity. Economics and Human Biology 3: 165–187.
Courtemanche C. 2009. Rising cigarette prices and rising obesity: coincidence or unintended consequence? Journal of
Health Economics 28(4): 781–798.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec
WEIGHT‐LOSS DIETING BEHAVIOR: AN ECONOMIC ANALYSIS

Cutler DM, Glaeser EL, Shapiro JM. 2003. Why have Americans become more obese? The Journal of Economic
Perspectives 17(3): 93–118.
Dansinger ML, Gleason JA, Griffith JL, Selker HP, Schaefer EJ. 2005. Comparison of the Atkins, Ornish, Weight
Watchers, and Zone diets for weight loss and heart disease risk reduction: a randomized trial. Journal of the American
Medical Association 293: 43–53.
Dockner E, Feichtinger G. 1993. Cyclical consumption patterns and rational addiction. The American Economic Review
83(1): 256–263.
Easterlin RA. 2001. Income and happiness: towards a unified theory. The Economic Journal 111(473): 465– 484.
Eisenberg ME, Neumark‐Sztainer D, Story M, Perry C. 2005. The role of social norms and friends’ influences on unhealthy
weight‐control behaviors among adolescent girls. Social Science & Medicine 60(6): 1165–1173.
Elfhag K, Rossner S. 2005. Who succeeds in maintaining weight loss? A conceptual review of factors associated with
weight loss maintenance and weight regain. Obesity Reviews 6(1): 67– 85.
Goldfarb R, Leonard TC, Suranovic S. 2006. Alternative motivations for dieting. Eastern Economic Journal 32(1):
115–131.
Gruber J, Frakes M. 2006. Does falling smoking lead to rising obesity? Journal of Health Economics 25(2): 183–197.
Harris JA, Benedict FG. 1918. A biometric study of human basal metabolism. Proceedings of the National Academy of
Science 4(12): 370 –373.
Hoie LH, Bruusgaard D. 1999. Predictors of long‐term weight reduction in obese patients after initial very‐low‐calorie diet.
Advances in Therapy 16(6): 285–289.
Ikeda JP, Lyons P, Schwartzman F, Mitchell RA. 2004. Self‐reported dieting experiences of women with body mass
indexes of 30 or more. Journal of the American Dietetic Association 104(6): 972 –974.
Jason MF, Frisvold D, Tefft N. 2010. Can soft drink taxes reduce population weight? Contemporary Economic Policy
28(1): 23–35.
Laibson D. 1997. Golden eggs and hyperbolic discounting. Quarterly Journal of Economics 112(2): 443– 447.
Levy A. 2002. Rational eating: Can it lead to overweightness or underweightness? Journal of Health Economics 21(5):
887–899.
Mattson MP. 2010. Perspective: does brown fat protect against diseases of aging? Ageing Research Reviews 9(1):
69 –76.
Morita Y, Iwamoto I, Mizuma N, Kuwahata T, Matsuo T, Yoshinaga M, Douchi T. 2006. Precedence of the shift of body‐
fat distribution over the change in body composition after menopause. The Journal of Obstetrics and Gynaecology
Research 32(5): 513–516.
Nayga RM. 2008. Nutrition, obesity and health: policies and economic research challenges. European Review of
Agricultural Economics 35(3): 281–302.
Nedergaard J, Cannon B. 2010. The changed metabolic world with human brown adipose tissue: therapeutic visions. Cell
Metabolism 11(4): 268 –272.
Neumark‐Sztainer D, Guo J, Story M, Haines J, Eisenberg M. 2006. Obesity, disordered eating, and eating disorders in a
longitudinal study of adolescents: How do dieters fare 5 years later? Journal of the American Medical Association
106(4): 559–568.
Neumark‐Sztainer D, Story M, Hannan PJ, Perry CL, Irving LM. 2002. Weight‐related concerns and behaviors among
overweight and nonoverseight adolescents. Archives of Pediatrics & Adolescent Medicine 156: 171–178.
O’Donoghue T, Rabin M. 1999. Doing it now or later. The American Economic Review 89(1): 103–124.
Oswald AJ, Powdthavee N. 2007a. Book review feature: Two reviews of the challenge of affulence: Self‐control and well‐
being in the United States and Britain since 1950. The Economic Journal 117(521): F441–F454.
Oswald AJ, Powdthavee N. 2007b. Obesity, unhappiness, and the challenge of affluence: theory and evidence. IZA
Discussion Papers 2717, Institute for the Study of Labor (IZA).
Pellizzon MA, Buison AM, Jen KL. 2000. Short‐term weight cycling in aging female rats increases rate of weight gain but
not body fat content. International Journal of Obesity and Related Metabolic Disorders 24(2): 236 –245.
Philipson TJ, Posner RA. 1999. The long‐run growth in obesity as a function of technological change. NBER Working
Paper 7423, NBER.
Popkess‐Vawter S, Wendel S, Schmoll S, O’Connel K. 1998. Overeating, reversal theory and weight cycling. Western
Journal of Nursing Research 20(1): 67–83.
Rosin O. 2008. The economic causes of obesity: a survey. Journal of Economic Surveys 22(4): 617–647.
Smith TG. 2009. Reconciling psychology with economics: obesity, behavioral biology, and rational overeating. Journal of
Bioeconomics 11(3): 249–282.
Suranovic SM, Goldfarb RS, Leonard TC. 2003. An economic analysis of weight change, overeating and dieting.
Unpublished Working Paper. (Available from: http://www.princeton.edu/~tleonard/papers/Dieting.pdf)
Suranovic SM, Goldfarb RS. 2006. A behavioral model of cyclical dieting. Advances in Health Economics and Health
Services Research 17: 49–78.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec
O. ROSIN

Tchernof A, Poehlman ET. 1998. Effects of the menopause transition on body fatness and body fat distribution. Obesity
Research 6(3): 246 –254.
Toth MJ, Tchernof A, Sites CK, Poehlman ET. 2000. Menopause‐related changes in body fat distribution. Annals of the
New York Academy of Sciences 904: 502–506.
Walsh MF, Flynn TJ. 1995. A 54‐month evaluation of a popular very low calorie diet program. The Journal of Family
Practice 41(3): 231– 236.
Yaniv G, Rosin O, Tobol Y. 2009. Junk‐food, home cooking, physical activity and obesity: The effect of the fat tax and the
thin subsidy. Journal of Public Economics 93(5–6): 823 – 830.

Copyright © 2011 John Wiley & Sons, Ltd. Health Econ. (2011)
DOI: 10.1002/hec

You might also like