Hey, David. I sincerely understand your point because I thought about the same problem before. Now I get some answer but not necessarily the right one. Here to discuss with you.

May I ask you a question first: Do you believe your research model is absolutely realistic? I'm doing research with mathematical model. We make many assumption before establishing a model. Then the model can capture the main idea of the research problem but cannot reflect the whole real world. The same can be applied to the assumption of ceteris paribus. With this assumption, we may investigate the relationship between key variables, which could provide some qualitative (even quantitative) insights for practice.

Assumptions highlight the research question. This is what I got so far.

Please let me know your opinion and I will appreciate different idea very much.

Thanks a lot Yingshuai for the opinion shared. Your point is well noted and much appreciated. However, in as much as models are believed to be abstract representation of reality, we in most cases believe or think that we have included all necessary variables, which may actually not be a representative depiction of reality. And should we go ahead to have other unrealistic assumptions backing our already flawed models instead of creating room for variations (which is actually what happens in the environment we live in), then how representative to reality are our models and findings...we may use some variables in a model and hold unto the assumption that should all other variables stay unchanged, a unit increase in X will lead to this change in Y. Let me ask you "Does only a single variable change in our environment while the others remain constant"? No matter how independent the variables may be, they mostly are never constant when others change.

A mathematical model is just an abstraction, more or less close (or distant) to the reality (like a map is not a territory, but a useful abstraction; there are many maps for different needs). It is of course easier to think in a static world than in a dynamic one, before introducing possible changes. The methodology copies, to some extent, experiments in hard sciences, in which the environment is "under control" so that it is possible to introduce changes and measure changes in the original experiments. To some extent, this methodology makes sense, but is of course unable to capture the whole subtlety of the reality. On top of that, there is the obsession of economist to search for the equilibrium, which is static by definition, and to think in relation to this equilibrium. In fact, economists have difficulties to comprehend dynamic changes and historical evolution. The most common way for them to do so is in the form of a percentage (like an interest rate). There are more sophisticated means to do so, from the "destructive creation" of Schumpeter to the new approaches of "endogenous growth", but the problem remain the same. Institutional economists such as Douglass North have included historical perspective in their approach, but then are not always thinking in terms of "ceteris paribus". But here, I am moving a bit too far from the original question.

Amazing....thanks a lot Louis for the wonderful opinion shared. Your point is well noted and very much appreciated

Its a great treatise from Louis. I love it. Innocent

to understand small changes of variables holding others constants seem to stem from the "marginal revolution" in economics around 1900. The same can be said of the assumed tendency towards an equilibrium which is just that an assumption (i personally think the existence of warehouses clearly tells us that most economic processes are out of equilibrium).

This way of tackling economic phenomena is understandable given the limited math tools at that time. Better tools are available since several decades. I suggest to apply "system dynamics" (www.systemdynamics.org, read e.g. John Sterman, Business Dynamcis). There you can model dynamics systems without bothering too much on mathematics itself but on their useful application.

Thanks a lot Robert for the interesting point raised. Your point is well noted and much appreciated

I read a book "out of control" yesterday and suddenly you question came to my mind again. In your question, you simulate the problem as a network and every point in the network has connections with each other, while in the traditional research, we simulate the problem as an isolated model which is controllable and predictable. The 100% simulation of a network is complicated and impossible, but it does be a future research direction. As Louis said, some researchers have already started to work on it. In my opinion, the way to establish a model really depends on your research question, the key problem you want to solve.

One way to think about this is an impact evaluation setting. You want to know whether a specific intervention (say, a conditional cash transfer) has the desired impact. But since lots of other things may be happening at the time and influencing the recipients' actions, you want to strip out those outside influences (say, via RCT or a natural experiment) and find out whether, ceteris paribus, the intervention is effective. If you cannot make that determination then you cannot attribute the impact to the intervention and would have a more difficult time making a policy decision based on these results.

thanks a lot Zhao and Denis for the wonderful opinions shared. Your respective points are well noted and much appreciated

Denis makes sense to me.

In economic models, there are endogenous variables, and exogenous variables. The model is in essence a theory about how the endogenous variables behave. The theory is based on the view that the exogenous variables determine the endogenous variables. Consumption behaviour of a group of people, for example, is determined by their incomes, the prices of the consumption goods they buy, and environmental factors like the weather that are influence their consumption behaviour. A natural question to ask in a model of this sort is this: how would a change in one exogenous variable affect the endogenous variables. The natural way to answer such questions is to change the exogenous variable of interest, holding other exogenous variables constant, and then observe what the model predicts with respect to the endogenous variables. In other words, the ceteris paribus assumption is what one needs to interrogate or explore a theory in which changes in certain variables are thought to be caused by changes in other variables.

thanks a lot Eaton for the wonderful opinion shared. Your point is well noted and very much appreciaed

Perhaps the notions of open an closed systems are more relevant to the question. Specifying the boundary conditions of the system, or rather, making assumptions, becomes inevitable if relationships between variables are to be investigated. Some economists choose a level of analysis that is relatively stable to circumvent this problem. But of course, any findings are valid only under those very conditions (assumptions).

This is indeed a question that is becoming more recurrent in academic circles. However, Ceteris Paribus continues to be used (all other things being equal) for a number of reasons. I guess one of them might be the low intellectual intensity of qualitative research, which is gaining momentum, and perhaps soon, qualitative research will gain more scientific affluence. Also, the recent rise of Heterodox economic represents, in my view, an intellectual counter-current that is certainly a reflection of discontent with the longstanding dominance of the quantitative paradigm in economics.

B. Eaton pretty much nails it. Many of the above comments bring in irrelevant opinions about the validity of marginalism, equilibrium, rationality assumptions, etc. But comparative "statics" and ceteris paribus assumptions are independent of all of these and in fact occur throughout the sciences. That's why we have partial derivatives. In economics they've also developed monotone comparative statics, which enables one in some cases to dispense with differentiability conditions and still draw conclusions.

Remember that comparative statics theory exercises are not intended to describe temporal differences--before and after--but are intended to describe logical alternative differences--if we were here rather than there. These theory exercises can encompass entire dynamic paths. (When climate scientists say that an increase in atmospheric CO2 will raise the average temperature in a climate model, they mean that as a ceteris paribus statement holding constant aerosols, solar output, etc., and they extend that result to time series of data, not just a single moment. They try to distinguish between transient and equilibrium effects, but both are ceteris paribus arguments.)

Thanks a lot Mohamed and Steven for the respective opinions shared. Your points are well noted and much appreciated

I am not certain how much I can add to the answers provided by Denis Medvedev, B. Eaton, Steven Postrel, and others here; however, that will not stop me from trying.

Although David Boansi's seemingly questions the use of the ceteris paribus assumption, it goes much deeper. At its core, the issue is mostly about epistemology and our scholarly process in economics. In most of the sciences, we strive to isolate and focus on one change at a time in an attempt to understand or analyze the impact of something. It is far more challenging to be definitive about functional relationships when multiple things change simultaneously (as they do in reality), .

All of us have probably encountered students who cite rising prices for fuel and increased consumption of it; they often ignore increasing affluence or the number of consumers in other countries. Instead, we strive to get them to offer conjectures that are both accurate and can be generalized. Most of the time it is challenging to offer something that is capable of being generalized without extremely narrow focus.

Like others have intimated, we use this as a scholarly process and along with it we use terms like ceteris paribus, partial derivatives, and comparative statics. May I suggest in all of this we are explicitly reminding ourselves and our students of what we are doing in this scholarly process?

In our attempt to be explicit in this scholarly process we also are (implicitly) trying to transform our students into scholars in the making. We hope (perhaps latently) they will be scholarly as they use this focus in their academic and professional endeavors.

I am convinced the core of ‎analytical thinking is focus and this is an attempt to improve understanding and knowledge. However, as David Boansi implies with his question, this involves a cost; we may loose sight of a bigger, more robust, or accurate picture. Just assembling fragments of knowledge may provide a distorted perspective of reality. That is a challenge where our success is questionable.

In fact, I think it is because of the point raised- social & economic situations are not always stable (constant) - that is why the maxim 'other things being equal' is used.

Otherwise, it will be helplessly difficult to make any analysis. In this case, we say there are other issues but am focusing on this particular one for now.

It is like a medical doctor treating a patient with multiple ailments - the dentist will face the aspect of teeth and leave other ailments for others specialist to handle.

Regards

In summation, ceteris paribus can be viewed as the verbal equivalent of a partial derivative.

Because you would be unable to report on any relationship if you assume that all is constantly changing.