Endogenous and ExogenousEconomists (particularly some conventional economists) quite often use the terms endogenous and exogenous variables when referring to economic models (and may extend them to real-world economic time series).
- An endogenous variable is a variable that is determined by the equations that define the economic model. (The endo- prefix is a Greek root that refers to something being internal.)
- An exogenous variable is a variable that is set by the user of the mathematical model, and it can be set to values independently of the operation of the equations of the model. (The exo- prefix is more familiar, and refers to something external.) Once all exogenous variables are fixed, it is possible to solve the model (do a simulation of a model run).
Under normal circumstances, exogenous variables have to reflect the decisions of one actor in the model, or possibly "laws of nature" (parameters in a production function). If there are two or more actors involved in determining the variable, the variable should normally be determined by looking at the equations that define the interactions; hence an endogenous variable. For example, the number of workers being employed is a joint decision between the worker to take a job, and an employer to offer the position, and so it would normally be endogenous. That is, it would be unusual to believe that the level of employment in the economy can be set by some external actor or law of nature.
(Some models will force variables that are best thought of as endogenous to be externally set for simplicity.)
A policy variable is a variable that theoretically could be fixed as an exogenous variable by a policymaking actor (the government), but it is set based on known conditions in the economy.
Under current institution arrangements, the best-known example is that of the short-term policy rate. It would be possible for a central bank to lock the policy rate at a fixed value (e.g., 0%) forever, but the current practice is for a committee to set the rate at periodic meetings. (There were historical periods where policy rates were essentially fixed, such as in the United States during and immediately after World War II.) It should be noted that even if the policy rate is fixed, there may be traded risk-free instruments whose interest rate may depart from those targets.
Within an economic model, the means by which the central bank sets the interest rate is its reaction function -- a term that is also sometimes applied to real-world central banks. The reaction function converts the exogenous interest rate into an endogenous variable. However, if one changes the reaction function, the model solution will presumably change.
Policy Variable Assignment: MMT versus RBC
I will compare the Monetary Monopoly model to the real business cycle (RBC) model found in Chapter 2 of Galí's introductory textbook* on dynamic stochastic general equilibrium (DSGE) models. Since the DSGE literature is vast, one could find many different frameworks, but my argument is that the RBC model chosen is representative of many benchmark models.
- The RBC model sets fiscal policy as a pair of exogenous variables: the real quantity of government consumption (which only appears in the sticky price model in chapter 3) and lump sum taxes. The Monetary Monopoly model uses lump sum taxes as well, but government purchases are specified in terms of the government setting a requisition price for some (aggregate) good, and purchasing as much as the private sector will provide.
- The RBC model discusses various interest rate rules for the central bank. As argued by Galí on page 21, the choice of an exogenous (fixed) interest rate results in price level indeterminacy. (DSGE model solutions are based on the expected value of variables in the future, and so the reaction function is forward-looking.) The Monetary Monopoly model does not allow for government debt, and so the risk-free rate of interest is set at 0% for all time. (This matches the policy preferences of at least some proponents of MMT.)
We see a stark contrast between the choices implied by the Monetary Monopoly model and the pre-2008 consensus among mainstream economists. That earlier consensus emphasised monetary policy, leaving fiscal policy as largely unimportant for stabilisation. Once interest rates hit 0% (and even negative in some regions), the consensus appears to have shifted back towards a mix of of fiscal and monetary policy. Although this is now closer to the MMT position, there is still disagreement about the effectiveness of positive interest rates to steer the economy.
Nevertheless, the use of price at which the government requisitions goods and services remains a key divider between conventional approaches (including post-Keynesian) and Modern Monetary Theory. It would presumably be possible to allow for this possibility within conventional models (such as DSGE models), but such an approach would be far from a consensus approach.
Although Modern Monetary Theory is part of the post-Keynesian school. the area of foreign exchange is where there are theoretical and policy disagreements. The MMT view is that the value of the currency should be left to float, as otherwise, the government is demoted to the status of "currency user," and faces constraints on fiscal policy. However, some post-Keynesians point to national accounting identities and argue that there is some form of "external constraint" even on floating currency sovereigns.
If the currency is pegged, the value of the currency unit becomes a policy variable. However, it can only be held at a target value so long as the government (central bank) has the capacity to intervene, and it needs to hold instruments with a stable value in foreign currencies (typically foreign exchange reserves, or gold in an earlier era) to do so. Once those reserves run out, the central bank is forced to abandon its peg (or devalue). As such, the exchange rate cannot be a pure exogenous or policy variable, rather some notion of being able to survive a run on the currency needs to be modeled.
The emphasis on the price of government requisitions (particularly the Job Guarantee), and the de-emphasis of monetary policy are the key divides between MMT and the consensus conventional position. (If we looked at the pre-2008 literature, fiscal policy was emphasised more by MMT, but the conventional consensus has arguably shifted since then.) Another policy variable choice -- whether or not to peg the currency -- is a major divide between MMT and other schools within the broad post-Keynesian tradition.
Link to final article in this sequence: Job Guarantee as a policy variable.
Link to final article in this sequence: Job Guarantee as a policy variable.
* Galí, Jordi. Monetary policy, inflation, and the business cycle: an introduction to the new Keynesian framework and its applications. Princeton University Press, 2015.
(c) Brian Romanchuk 2020
“The emphasis on the price of government requisitions (particularly the Job Guarantee), and the de-emphasis of monetary policy are the key divides between MMT and the consensus conventional position.”ReplyDelete
When you say “particularly”, I can see how the JG lends itself naturally to a characterization of price setting. But what other practical examples of government spending would be similar to this?
Presumably a government with the JG still spends money on various things other than wages paid for services rendered in the JG. How does price setting apply outside of the scope of the JG?
Wouldn’t the full spending menu even in the case of a JG involve a mix of price setting and price taking?
And if so, don’t the proportions of each matter in the full characterization?
Not sure if these questions make sense – just thinking about it now.
The JG is the best example, and I will be discussing it next (last of the planned sequence of articles).Delete
As for other examples, the most important would be wages of regular employees. Although this is normally thought of as price taking, wages are administered (typically with union negotiations). Another area is that of transfer payments (admittedly not a requisition price, but can be thought of as having a similar effect).
Requisition of goods/services from the business sector is typically via auction, and by definition is price taking. The obvious point was that this behaviour is not a law of nature, and could be viewed as a policy error.
One can say that governments do not think this way. The response is that is one major reason why inflation took hold in the 1960-1970s. I will immediately note that is my initial thinking on the topic, and I do not recall anyone arguing that point specifically.
The MMT argument is that budgeting is to be done with an eye on inflation. Even if the government is not attempting to dictate prices (which is going to be hard to do in many cases), the budgeting should take price changes into account.
In the current environment of “low-flation”, observed CPI is going to mainly follow sectoral effects in the private sector. These wiggles are not going to be related to what the government is doing. However, if we look at the US, the biggest engines of inflation in the past couple of decades has been higher education and medical care, which are almost entirely due to government policy. Rent is another area of inflation, and is also related to policy at differing levels of government.
Another area is indexation. Indexation is obviously a concern when looking at inflation control, and it does affect prices and wages paid by the government.Delete
(Ideas still coming...)Delete
Typical government budgeting puts everything under nominal dollar envelopes. This creates a hybrid system, where the quantity demanded is a function of price. This helps deal with the theoretical issue of instantaneous jumps in the spot price level that pop up in economic theory (but not so much in the real world).
This comment has been removed by a blog administrator.ReplyDelete