What is the definition of exogenous and endogenous preferences?

Wikipedia states:

Exogenous Preference -- one that comes from outside the model and is unexplained by the model.

Endogenous Preference -- preferences then cannot be taken as given, but are affected by individual internal responses to the external state of affairs.

Can someone clarify what these two terms mean? Specifically, what does it mean to be "inside" (or internal) or "outside" (or external) of a model?

• What wikipedia article is that? It could use some improvement. Apr 28 '15 at 21:40
• I'm not sure I got this proof quite right. economics.stackexchange.com/a/5383/2679 It's an answer to my own question, but I think I may have worded the part about the inverses wrong. Specifically, I know you can have an ID curve that fails strict convexity and is tangent to the budget line at two points. So even though $v$ and $e$ would be inverses, it's incorrect to say there is only one bundle for which they act as inverses. But I don't know how to say this rigorously / formally. Apr 30 '15 at 1:22

Exogenous variables are believed to have some value given by nature. They are not caused by your theory's variables of interest. This is why they are said to be outside the model.

Endogenous variables have values dependent on your theory's variables of interest. They both cause, and are caused by your topic.

Example: In the study of wages, some individuals are born superior, genetically, perhaps. They are superior exogenously. Because of their superior innate quality, they may choose to seek additional education knowing they will get higher returns for their investments. So the amount of education is endogenous to wages, because a person's expected earnings informs their education.

The inside/internal/endogenous thing simply means that you have something inside your model that determines the value of this specific value.

To put it differently: the model structure (e.g. its underlying equation) + the exogenous variables you throw into your model determine the endogenous variable.

An Exogenous Variable is defined as a variable which is unaffected by other variables within an model.

take a multivariable regression model as an example:

$$y=\beta_0+\beta_1x_1+\beta_2x_2+u$$

$x_1$ is called an exogenous variable when its determination is unaffected by $x_2$ and error term $u$.

An Endogenous Variable is defined when $x_1$ is influneced by $x_2$ or $u$.

This is important because when we run a regression we are producing a function which assumes a dependent and independent variable. If we find endogeneity we wont get accurate estimates for the effect of $x_1$ on $y$.

Thus, using endogenous preferences allow a researcher to study phenomena where social interaction (e.g. networks, influence, marketing, etc) is relevant for consumption choices. For example, here is a paper studying "fashion cycles". Preferences for individual $i$ are quite simply define as Cobb-Douglas functions:
$$U^i = x^{\alpha_i} y^{1-\alpha_i}$$
where $x$ and $y$ are two goods, and $\alpha_i \in (0,1)$. Preferences are endogenised by assuming that $\alpha_i$ depends on society's past level of consumption of good $x$. Therefore, how much other individual have consumed in the past affects how much an individual would like to consume today. In this sense, preferences are endogenous. This contrast with common exogenous preferences where individual choices do not depend on other individual choices (other than through common market mechanisms like price, demand-supply).