What’s in a model… (an economic one, that is)

What’s an economic model? A little semantic intro (don’t worry, we will get to Keynes in a moment)

I work together with biologists, agronomists and even test animal specialists on a regular basis. These people use words like ‘conceptual models’ or ‘even ‘animal model’ all the time. Check this:  ‘Mouse Models of Diabetic Nephropathy‘. Yes, a live animal is considered to be a scientific model. I had to get used to this as this use of the word ‘model’ transcended the boundaries used in the world I came from. But mastering these concepts prooved enightening. Conceptual models are described by Andrew Powell-Morse in the following way:

A conceptual model is a representation of a system that uses concepts and ideas to form said representation… a model is intrinsically a thing unto itself, but that model also contains a concept of what that model represents — what a model is, as opposed to what a model represents.  (think of the mouse and diabetes, M.K.) … conceptual modeling is used as a way to describe physical or social aspects of the world in an abstract way. For example, in the realm of software development, a conceptual model may be used to represent the relationships of entities within a database. Whether written down via text or diagrammed visually, such a conceptual model can easily represent abstract concepts of the relationships between objects in the system, such as Users and their relationship to Accounts.

Did I mention that monetary transactions like investments or wages paid are relationships between abstract concepts like ‘household’ and ‘the government’ and ‘non-financial companies and the like which are estimated using the national accounts? Well, they are. Which brings us to Keynes and the national accounts. A conceptually pathbreaking article of Keynes was ‘How to pay for the war’. For the first time, as far as I know, an estimated system of national accounts with different income groups (low-income and high income) was used for policy analysis and prescriptions.  Keynes’ World War I experience had taught him the difference between profit inflation and wage inflation. Price increases during WWI were mainly caused by increasing profit margins. Good for ‘capitalists’ (or the high income class in ‘How to pay for the war’), bad for ‘labor’ (the low income class). He wanted England to be able to free resources for the War Effort without unwanted social consequences. Keynes was keenly aware that simply raising government spending would do the trick, but at the price of profit inflation and hence suffering of the poor without suffering of the rich. A nasty twist to this: Utsa Patnaik shows that the 1942/1943 famine in Bengal, at that time part of the English colony India, was caused by a deliberate English policy of profit inflation. He also shows that in all probability Keynes was at least keenly aware of this. The point: the way we measure the macro economy and the concepts we use to do this matters. A lot. It’s sometimes about life and death.

Why all this? Because of a little twitter fitty I had with the esteemed @Noahpinion. I do not use ‘esteemed’ in an ironic sense. He’s very well read, has very informed opinions and writes very well. I learn a lot from him. More importantly: he is really expanding the realm of data based economic discussion – something I only dream off. This time he however stated: “accounting is not a model of the economy“. Which is wrong. The national accounts are a model of the economy. And economists have to learn it is. The way the entities are conceptually defined matters as this also defines the monetary relations we see. Concepts and definitions matter. What is included, or not, in ‘financial institutions’ or ‘investments’ is conceptual and an important choice. I, for one, think that artificial hips should be included in the definition of investments and fixed capital as these have a monetary production cost as well as a monetary price while they do know some kind of depreciation. The accounts (together with the manuals, I should say) are a conceptual model of our economy (think of the mouse and diabetes again). And choices are made how to do this. Examination question: why was Keynes’ highly useful, estimated and influential conceptual innovation not incorporated in the post WWII national accounts?

Noah Smith links to a Brad De Long blogpost which states (while approvingly mentioning Paul Krugman and Noah and Mark Thomas):

it turns out in economics to be remarkably hard for lots of people to distinguish between:

• behavioral relationships–things that tell you how people will change their behavior to respond to changes in the economic environment and economic policy;
• equilibrium conditions–things that tell you what configurations of the economic environment are consistent and are not rapidly-changing out-of-equilibrium phenomena seen for an eyeblink of time, if that long; and
• accounting identities–things true by the metaphysical necessity of the definitions that are devoid of interesting substantive implications.

Brad is totally right that we should distinguish these. But he is wrong about the accounting identities: the accounting identities we measure are dependent on the way the national accounts model the economy. Also they are not metaphysical. They are a and emergent consequence of the legal and social and cultural aspects of money and monetary contracts. You do not need all three aspects for a model of the economy.  To come back to ‘How to pay for the war’: all of these three items were present albeit sometimes in a rather implicit way. But the ‘definitions’ were not devoid of ‘interesting substantive implications’, the conceptual difference between profit and wage inflation in relation to a delineation of different groups of households was crucial (Keynes being Keynes the income groups were already measured, See this article of E. Rothbarth). To do this, Rothbarth used guided by Keynes a conceptual model of the economy with multiple income groups. National accounts do use a model of the economy, the accounting identities are based on the fundamental social properties of money and monetary transactions but for the way we measure them a conceptual model is key.

Why this rant? I’m having the idea that in the MMT discussion some economists have trouble with the definition of ‘model’. They have to be more precise when using this word. Alas, a conspicuous use of fuzzy definitions is, as Veblen already noted in 1909, not a bug but a feature of neoclassical economics. Look at this model by Bokan e.a., which does not know if it has to measure labor in hours or persons and uses undefined ‘labour services’ instead. We can do better than that.