UNDERSTANDING THE MOOD-DIETING CYCLE

IAREP-SABE 2009 Conference, Halifax, Canada May 2009 UNDERSTANDING THE MOOD-DIETING CYCLE BING JIANG Department of Economics Emory University, Atlan...
Author: Asher Gardner
1 downloads 2 Views 267KB Size
IAREP-SABE 2009 Conference, Halifax, Canada May 2009

UNDERSTANDING THE MOOD-DIETING CYCLE

BING JIANG Department of Economics Emory University, Atlanta, GA 30322, U.S.A Email: [email protected] March, 2009

Abstract This paper develops a behavioral model in food consumption choices that illuminates observed patterns of weight gain and dieting. Particularly it addresses the important role of willpower and mood stability in dieting behavior and hence consumption decisions. Built on previous research of the “dual-process” model (see Jonathan, 2005 and Loewenstein & O’Donoghue, 2005) in which a person’s behavior is the outcome of an interaction between deliberative process and affective process, and Baumeister and Vohs (2003) that attempts by the deliberative system to override affective motivations require limited supply of willpower, this paper further shows that people cannot exert full control over their goal-pursuing behavior, as reflected by the existence and fluctuation of moods during the pursuit. What is new to research is, this model assumes mood status is an important variable affecting willpower to exert control in dieting behavior, i.e. a good mood helps people diet successfully, whereas a bad mood weakens their willpower of fulfilling dieting, so dieters gain excess weight. In designing the experiment, we take into account the subjects’ emotional experience during dieting, and use Positive affect and negative affect schedules (PANAS) developed by several psychologists Watson, et al. (1988) to record their mood status. Indeed, there exists a noticeable amount of evidence in psychological studies supporting that good mood, or positive affect increases intrinsic motivation of achieving a goal and ability of self-regulation. Our experiment comprises of university students who volunteer to try diet food to lose weight. Everyday they are provided with a certain amount of diet food and requested to report how many calories as well as how many other types of food they actually consume on that day. PANAS are administered to record their mood status during and after each experimental manipulation. After the whole experimental session, their weight gain/loss will be measured. We hypothesize that, as subjects’ moods fluctuate during dieting, their willpower of exerting control over non-diet food has been undermined. Moreover, abstention results in stress, anxiety and depression, which make mood fluctuate more. Subjects’ willpower in dieting is further undermined. This eventually leads them to subsequent overeating behavior and continuous dieting cycle, known as the “mood-dieting cycle”.

1

IAREP-SABE 2009 Conference, Halifax, Canada May 2009

1. Introduction In recent years the problem of increasing obesity in the U.S has sparked research interest across a variety of disciplines. According to Mokdad (2001), fully 35 percent of Americans are overweight, and an additional 20 percent are obese. Many overweight people prefer not to be overweight, as the existence of a large diet industry suggests. Americans pay $40-$100 billion annually to help them lose weight. The Wall Street Journal (Parker-Pope 2003, p. R-1) reported that “at any time, 20% of men and 44% of women are on a diet.” Recent economics literature on this topic has attempted to empirically identify the causes of the observed rise in obesity over time (e.g., Chou et al., 2004; Lakdawalla and Philipson, 2002). In addition, there has been worth to develop behavioral models in self-control that explain food consumption choices leading to excess weight and dieting (e.g., Levy, 2002; Dockner and Feichtinger, 1993). There are both practical and analytical reasons for paying particular attention to dieting behavior. On the practical side, dieting is extensive and involves significant expenditures of effort and money. On the analytical side, dieting behavior presents a number of conceptual challenges for rational choice models of human behavior. This research contributes to the discussion of behavioral models of self-control by developing a model of mood self-regulation in food consumption choices that illuminate observed patterns of weight gain/loss and dieting. It argues that mood status significantly affects the willpower of fulfilling dieting and hence, food consumption decisions.

2

IAREP-SABE 2009 Conference, Halifax, Canada May 2009

2. Literature Review Why do people have self-control problems? Hoch and Loewenstein (1991) provide a possible explanation, known as time-inconsistent preferences. According to this analysis, consumer decisions represent an ever-shifting conflict between desire and willpower. When the desire to purchase a product outstrips consumers’ intentions not to make the purchase, impulse buying can result. Their research highlights two separate mechanisms that determine consumption decisions: (1) the desire to buy and (2) the ability to exercise control over this urge. In two recent articles, Jonathan (2005) and Loewenstein and O’Donoghue (2005) develop a dual-process model in which a person’s behavior is the outcome of an interaction between deliberative process and affective process. Deliberative process assesses options with a broad, goal-based perspective. The affective process encompasses emotions such as anger, fear and motivational drives. In the dual-process model the affective system has initial control, but the deliberative system can influence behavior through the exertion of effort, or willpower. Whether or not the deliberative system dominates behavior depends on the trade off between its objectives against the cost of exerting this willpower. That means, willpower is a resource expended by deliberative process to wrest influence from affective process. Other research supports many common intuitions about willpower. Baumeister and Vohs (2003) suggest that attempts by the deliberative system to override affective motivations require an inner exertion of effort, often referred to as “willpower.” They argue that willpower is in limited supply, at least in the short term, and exerting willpower in one situation tends to undermine people’s propensity to use it in a

3

IAREP-SABE 2009 Conference, Halifax, Canada May 2009 subsequent situation. This paper builds on previous research by developing a model where willpower is limited and bounded, as reflected by the existence and fluctuation of moods in constrained optimization problems. Bounded willpower means that people cannot exert full control over their goal-pursuing behavior. For those who are on a diet, bounded willpower and mood effects often lead to relapse by abstinent addicts into other non-diet foods. What is new to research is, this model assumes that mood status is an important variable affecting willpower to exert control in dieting behavior, e.g. a good mood helps people diet successfully, whereas a bad mood weakens their willpower of fulfilling dieting, so dieters gain excess weight.

3. Model and Hypothesis 3.1 The Model According to “bounded rationality” choice model, individuals do consider the future health effects of today’s diet, but do not consider how today’s diet will affect tomorrow’s diet decisions. Therefore, an individual maximizes utility in each period rather than projecting the long-term impact of food on weight and future utility. Alternatively Constantinides (1990) proposes a habit-formation model in which the utility of consumption is assumed to depend on past levels of consumption. Specifically, consumers are assumed to be averse to reductions in their level of consumption, as Kahneman and Tversky (1979) suggest in their prospect theory where utility is defined over gains and losses (i.e., returns) rather than levels of wealth.

4

IAREP-SABE 2009 Conference, Halifax, Canada May 2009 Hence, the model in this paper combines “bounded rationality” and habit formation models, and assumes that individuals update their utility level as a result of choices made in each period based on contemporaneous mood status. To simplify the problem, it involves only tth period, consumption of two types of food: diet (healthy lowfat) food Xdt, and regular (potentially fatty) food Xrt , and mood status, mt in period t. So for a representative dieting agent, named Sophie, her utility at time t is given by: Ut =Vt (Xdt, Xrt, mt),

(1)

where mt is a function of previous mood status mt-1, current utility level Ut and other exogenous factor (s) et that affect mood at period t. Hence, mt = ρmt-1 + θUt +et, with ρ>0, θ>0.

(2)

Key assumptions of this model are: (3)

∂Vt (i) > 0, ∂X dt

∂Vt (i) > 0. ∂X rt

(4)

∂ 2Vt (i) < 0, ∂X dt 2

∂ 2Vt (i) < 0. ∂X rt 2

(5)

∂ 2Vt (i) > 0, ∂X dt ∂mt

∂ 2Vt (i) < 0. ∂X rt ∂mt

Assumption (5) implies that a good (bad) mood in dieting raises (reduces) the marginal utility of consuming diet food and reduces (raises) the marginal utility of consuming regular food. This means that if Sophie has a good (bad) mood in period t, she will increase (reduce) the consumption of diet food and reduce (increase) the consumption of regular food. Through assumption (5), the relationship of how immediate mood status affects consumption choices and hence, utility level has been established. Given an initial mood status, m0, Sophie chooses {Xdt, Xrt} to maximize the

5

IAREP-SABE 2009 Conference, Halifax, Canada May 2009 discounted utility flow Ut, subject to the budget constraint PdtXdt +PrtXdt ≤ It. This can be written as: Max Ut =Vt (Xdt, Xrt, mt)

s.t.

{Xdt, Xrt}

PdtXdt +PrtXdt ≤ It ,

(6)

where Pdt and Prt are the prices of diet and regular foods, and It is Sophie’s income level at period t.

3.2 Hypothesis and Discussion 3.2.1

Determinant of mt and applicable implications

To see the underlying relationship between mt and Ut, plug mt-1 = ρmt-2 + θUt –1 + et-1 into equation (2). So equation (2) becomes: mt = ρ(ρmt-2 + θUt –1+ et-1 ) + θUt + et = ρ2 mt-2 + ρ θUt –1+ ρet-1+ θUt + et = ρ2 (ρmt-3 + θUt –2+ et-2) + ρ θUt –1+ ρ et-1+ θUt + et = ····· i = t −1

j = t −1

i =1

j =1

= ρt m0 + θ (ρ ∑ U i + Ut) + (ρ

∑e

j

+ et )

(7)

Therefore, mt is a function of initial mood status m0, sum of previous utility i = t −1

flows ∑ U i , current utility Ut , and the total effect of exogenous factors. Since assuming i =1

that both ρ and θ are positive, and Ut is an increasing function of mt, as shown from equation (2) and (3), one prediction of this model would be: if Sophie is free to maximize utility in any given period t, then her maximized utility in that period is greater when she starts the period in a better mood (m0 is higher) than when she starts the period in a worse mood. If this hypothesis holds, an interesting application of this model is to find out possible factors that may affect m0, one determinant of mt, to generate unexpected results

6

IAREP-SABE 2009 Conference, Halifax, Canada May 2009 with the knowledge of solving constrained optimization problems. From the above analysis, this model may be able to explain why poor people are usually fatter than rich people. If we assume that everything else is equal for poor and rich people, aside from their income levels, then initial mood status m0 could be different, i.e. poor people have a lower initial income than rich people, so poor people have a very low initial mood status m0. In this model, low m0 leads poor people to increase the consumption of regular food (MUrt is rising) and decrease the consumption of diet food (MUdt is falling). As a result, poor people who eat more regular yet potentially fatty food than diet and healthy food are fatter than rich people. This is one application of how mood status, as an important variable, can be embedded into a traditional constrained optimization model and generate economically interesting results. 3.2.2

Income Effect of Mood

In section 3.2.1, I have developed some applicable implications by extending mt and analyzing how initial mood status m0, one determinants of mt, affect consumption choices in dieting behavior. In this section, I will prove that with the influence of a mood variable, regular food (Xrt) tends to be inferior, i.e. people demand it less as their income increases. The proof shows that with everything else but income remaining unchanged, if income is rising (falling), mood is increasing (decreasing), then the representative agent, Sophie will tend to eat less (more) Xrt. First, let’s see the simple constrained optimization problem without mood: Max U(x, y)

s.t. px+qy= I,

(8)

{x, y}

where {x, y} is the consumption choice of good x and good y, p and q are prices of these two goods, and I is the income level. 7

IAREP-SABE 2009 Conference, Halifax, Canada May 2009 Writing y as the function of x, p, q and I, and plug the function of y into the object function, then (8) becomes: I p MaxU ( x, y ) = MaxU ( x, − x) { x , y} x q q ∂U (i) dy p F.O.C: = MU x + MU y i = MU x + MU y i(− ) = 0 dx q ∂x ∴

S .O.C :

(9) (10)

MU x p = MU y q

(11)

∂ 2U p p2 2 = U − U + U yy < 0 xx xy ∂x 2 q q2

(12)

Implicitly differentiate F.O.C, or equation (10), we get: p p U xx dx + U xy dy + (− )U yx dx + (− )U yy dy = 0 q q Plug dy =

(13)

dI p − dx into equation (13), it becomes: q q (

U xy



pU yy

)dI + (U xx −

p p p2 U yx − U yx + 2 U yy )dx = 0 q q q

q q2 U xy pU yy dx ∴( − 2 ) + ( S .O.C.) = 0 q q dI pU yy U xy p U yy U xy ( ) − − 2 dx q q q q p ∴ = = dI S .O.C. S .O.C

(14) p is positive, x is a q

Since S.O.C requires the denominator of (14) to be negative, and normal good if

U yy q




U xy p

=

U yx P

.

Now, adding the mood variable, denoted by mt, into the optimization problem, we have: Max Ut =Vt (Xdt, Xrt, mt)

{Xdt, Xrt}

where

s.t.

PdtXdt +PrtXdt ≤ It ,

mt = ρmt-1 + θUt +et, with ρ>0, θ>0.

(6) (2)

8

IAREP-SABE 2009 Conference, Halifax, Canada May 2009 To further simplify the problem, assume that previous mood mt-1 is exogenous, denoted by C, therefore, equation (2) can be written as mt = ρC + θUt +et

∂U t dx ∂U t = Vd dt + Vr + Vm iθ =0 dxrt ∂xrt ∂xrt P V ∴ r = rt Vd Pdt

∴ F .O.C :

(2)’ (15) (16)

∂ 2U t P dX dt Prt P dm t dX dt dm t S .O .C : = − rt V dd i − Vdr − rt Vdm i + V rr + Vrd i + V rm i 2 Pdt dX rt Pdt Pdt dX rt dX rt dX rt ∂ X rt +θ

dX dt dm t ∂U t ∂U t ∂U t ∂ 2U t V md i V mr + θ V mm i +θ + θ Vm i 0 and Vrm = , mood enforces this effect: as Pdt Prt Prt It becomes lower, people tend to eat more and more regular yet unhealthy food;



If Xrt originally is an inferior good, or



If Xrt originally is a normal good, or

Vdd U rd U dr < = , mood can make Xrt inferior. Pdt Prt Prt

4. Methodology In general analysis, this model is a constrained optimization problem and can be solved by applying usual optimization methods. The result shows that a good mood helps people diet successfully, while a bad mood weakens their willpower of fulfilling dieting, so dieters gain excess weight. Also, this model can explain how possible factors that affect initial mood status, e.g. initial income, may affect consumption choices of certain foods and maximized utility. The conclusion is based on the assumption that a person can maximize present utility and that the greater her initial mood status, the greater she can make her end-of-period utility. On the other hand, there are many experimental evidence showing that good mood, or positive affect increases intrinsic motivation of achieving a goal and ability of self-regulation. Since this paper intends to test the hypothesis that the mood/emotion status can affect individuals’ results of fulfilling dieting, conducting experiments is

10

IAREP-SABE 2009 Conference, Halifax, Canada May 2009 indispensable. The conceptual experimental design can be composed of university students who volunteer to try diet foods to lose weight. Everyday they are provided with a certain amount of diet food and report how many calories as well as how many other types of food they actually eat on that day. After the experimental period, their weight gain/loss will be measured. Secondly, in order to test how mood status relates to food intake decisions for individuals who are on diet, positive affect and negative affect schedules (PANAS) may be administered to record participants’ mood status during and after each experimental manipulation. Using the data from the experiments, a series of tests of proposed hypothesis may be established.

11

IAREP-SABE 2009 Conference, Halifax, Canada May 2009

Reference Baumeister, R. F. and D. V. Kathleen. “Willpower, Choice, and Self-Control,” in George Loewenstein, Daniel Read and Roy F. Baumeister, eds., Time and Decision: Economic and Psychological Perspectives on Intertemporal Choice, New York: Russell Sage Foundation, 2003, 201-216. Bernheim, B. D. and A. Rangel. (2004). “Addiction and Cue-Triggered Decision Processes.” The American Economic Review, 94 (5): 1558-1590. Chou, S., H. Saffer, and M. Grossman. (2004). “An Economic Analysis of Adult Obesity: Results from the Behavioral Risk Factor Surveillance System,” Journal of Health Economics, 23 (3): 566-588. Cohen, J. D. (1991). “The Vulcanization of the Human Brain: A Neural Perspective on Interactions between Cognition and Emotion and Optimality in Decision Making.” Working paper, Department of Psychology, Princeton University. Constantinides, G. M. (1990). “Habit Formation: A Resolution of the Equity Premium Puzzle.” The Journal of Political Economy, 98 (3): 519-543. Dockner, E. and G. Feichtinger. (1993). “Cyclical Consumption Patterns and Rational Addiction.” American Economic Review, 83 (1): 256-263. Hoch, S. J. and G. F. Loewenstein. (1991). “Time-Inconsistent Preferences and Consumer Self-Control”. The Journal of Consumer Research, 17 (4): 492-507. Kahneman, D. and A. Tversky. (1979). “Prospect Theory: An Analysis of Decision under Risk.” Econometrica, 47 (2): 263-292. Laibson, D. (1997). “Golden Eggs and Hyperbolic Discounting.” Quarterly Journal of Economics, 112: 443-477. Lakdawalla, D. and T. Philipson. (2002). “The Growth of Obesity and Technical Change: A Theoretical and Empirical Examinations” NBER Working Paper, 8946. Levy, A. (2002). “Rational Eating: Can it Lead to Overweightedness or Underweightedness?” Journal of Health Economics, 21: 887-899. Loewenstein, G. F. and T. O'Donoghue. (2004). "Animal Spirits: Affective and Deliberative Processes in Economic Behavior” SSRN working paper. Mokdad, A. H., B. A. Bowman, E. S. Ford et al. (2001). “The Continuing Epidemics of Obesity and Diabetes in the United States,” JAMA, 286 (10): 1196-1200. Parker-Pope, T. (2003). “The Diet that Works,” The Wall Street Journal, April 22: R1. 12