# Stock-Flow Consistent models

Stock-Flow Consistent (SFC) models are becoming more popular among heterodox economists (particularly in the Post-Keynesian school of thought). But I fail to see the relevance of this approach. The main addition seems to be the "construction" of the "Flow of Funds" table. But is that really necessary? Whenever an economist is doing a model, careful attention must be presented to including relevant sectors, and trying not to have "inflows/outflows" which are not explained in the system.

What is so special/new about SFC models? Are not most of mainstream models SFC? Even more, what if a model is not SFC, but is a better predictor/more realistic than one with SFC?

• Very good question, I am interested in the answer too ! I heard most macro models are SFC (including dsge), i am not sure though. Post Keynesians emphasise money creation through credit in their models, so stock flow consistency seems to be both more important and less trivial to obtain there – Ululo Mar 7 '17 at 21:05
• @Ululo Apparently, DSGE models are not SFC. That is what a guy said in a recent SFC workshop I went, trashing everything about DSGE, but he did not back his claims with evidence/test/examples. – luchonacho Mar 8 '17 at 9:48
• I am not specialist and its been a while: in a seminar I went, a guy was very excited by SFC models since it was allowing him to study all sort of network-like redistributive questions. His "flow of funds table", say, $A$, was square, and he was able to study quite precisely the impact of a given tax on different types of shareholders, (indirectly by studying how the tax flows over different sectors) using the Leontieff inverse $(I-A)^{-1}$. – keepAlive Mar 9 '17 at 0:25
• My understanding (that is not enough for writing a full answer) is that macroeconomics stock-flow models make obvious relationships like "government spending surplus equals non-government sector deficit" that are not fully acknowledged by mainstream economics. A reading that help me to understand it: bilbo.economicoutlook.net/blog/?p=4870 – Robert Mar 9 '17 at 17:18

Marc Lavoie, page 265 of Post-Keynesian Economics: New Foundations: "In fact, the SFC approach is perhaps a misnomer, as several other theories relate stocks and flows in a consistent way. The peculiarity of the post-Keynesian SFC aporoach is that its models truly integrate the real and monetary sides."

In other words, do not focus on the name "stock-flow consistent models," the (alleged) contributions go beyond stock-flow consistency.

The usefulness of SFC models depends upon your theoretical perspective.

If you are already a post-Keynesian economist, SFC models appear to offer a rigourous, standardised framework for setting up economic models. One of the problems facing post-Keynesian approaches is that there are considerable differences of opinion in how to approach economics, and the mathematical models used appear quite different.

The hope was that some of the distinctions might disappear if cast into the same mathematical framework. In practice, there remain considerable differences in the behavioural equations between various post-Keynesian schools of thought, so SFC models have not been able to completely unite the various schools of thought. (The text by Lavoie discusses this at greater length.)

If you are not post-Keynesian, the usefulness depends upon your views about mainstream macroeconomic models, which I will summarise as dynamic stochastic general equilibrium (DSGE) models here as a shorthand. Please note that there is a massive number of variants of DSGE modelling; my comments here reflect the views of an external observer who is not worried about the distinctions between those models.

(Since SFC models have an explicit economic model structure, they are not comparable to unstructured models that are generated by a statistical fitting procedurs.)

From a post-Keynesian perspective, DSGE models embed a great many assumptions, pretty much all of which have been criticised. For example, utility maximisation, the production function, rational expectations, general equilibrium, representative households, price determination, and even the distinction between randomness and uncertainty have all come under attack by post-Keynesians. (Since "DSGE models" vary, not all of these assumptions appear in all models. However, it is probably safe to say that almost all "DSGE models" incorporate at least some assumptions that at least some post-Keynesians object to.) SFC models do not embed those assumptions, and so have an advantage if you view those assumptions as incorrect.

However, even if you move past the arguments over the assumptions, the tractability of SFC models is an advantage versus pure DSGE models. Standard DSGE models have to make simplifying assumptions in order to remain tractable, such as not having a financial sector. Although certain DSGE models can relax such assumptions, what typically happens is that only some of them can be relaxed at once.

In practice, many central banks have had to use hybrid models, that only incorporate some behavioural equations taken from DSGE theory. Although it might not meet a purist's definition of a "SFC model", you could easily create such a hybrid model from a SFC model base. From the perspective of applied mathematics, I would be hard-pressed to give a formal mathematical reason why such a model is not a member of an extended family of "SFC models." (It would not be classified as a DSGE model, as there is no assumption of equilibrium. The behavioural equation is just a bolted on assumption, independent of equilibrium arguments.)

However, if you insist that macro models must reflect the optimising choices of households, SFC models are not going to help you very much.

• It seems you are using SFC and Post-Keynesian as equivalent, which contradicts Lavoie's quote. Without defending DSGE, there seems to be plenty of variation within that modeling framework, departing from some of the assumptions you list. Some of them are not necessary nor sufficient for a model to be DSGE, just as not having them means the model is SFC by definition. – luchonacho May 15 '17 at 16:24
• My interpretation of the Lavoie quote is that the name SFC models does not imply that other approaches (including mainstream) are not stock-flow consistent. As I indicated, SFC models were not able to unify all PK approaches. As for DSGE models, yes they are varied, but pretty much all of them get hit by some PK critique. However, it seems that they need at least an equilibrium assumption to qualify for the "GE" in "DSGE." Anyway, I will make some modifications to try to address this comment. – Brian Romanchuk May 15 '17 at 17:37