I'm not a huge macro guy. This statement can be confirmed by either my classmates or my grades in macroeconomic courses, (undergrad A/B-, grad school C+/B). But I think my general ignorance to this aspect of the "dismal science" can actually help in providing a decent blog as to why the 2011 Nobel Prize winners in Economics, Thomas Sargent and Christopher Sims were so deserving of the honor. Reading the press reports confirms that reporters haven't the foggiest clue as to why Sargent and Sims' contributions were gigantic...
UChicago news here. (we are always quick to claim a Nobel as our own but this is really a Minnesota coup)
Excellent but probably inaccessible write up on Marginal Revolution by Tyler Cowen: Sims here and Sargent here. Alex Tabarrok's words here.
Probably the best layman read is on the nobel.org webite here. They provide a good 5 page explanation with pretty pictures. But even that would be difficult without having recently taken a macroeconomic and econometric course. Why is macro so opaque?
First, macro is complicated stuff and we know so little about the way large scale economies work. This is funny because macroeconomists, central bankers, and IMF types wear power suits and look like people in charge of the world. They speak well, look good, and use big words in a commanding voice but beneath the Brooks Brothers tie they know deep in their hearts that their models have little predictive capability. This isn't to say they know nothing or are useless it is just that the complexity of the field is high. In contrast the typical microeconomist looks disheveled and like somebody you should help into a shelter not an expert you should ask for advice about economics.
Second, nothing seems to be settled in the field regarding an overarching structure. We've gone round and round with Keynes, Hayek/Friedman, New Keynsian, and we are coming back to the classical models regarding our fiscal policies like real-business cycles. The variability in macro ideas between DeLong, Krugman, Romer, Summers, Mankiw, the Chicago school, etc....is enormous. In contrast I can't think of a microeconomic policy debate where people are that far apart. Wearing an economist hat, I'm fairly agnostic about which macro models we should bet on. Ideologically, color me Austrian.
Thirdly, when the core of your field rests on an definition/equation
GDP=Consumption + Investment + Government Spending + Exports - Imports, you've got huge inadequacies to deal with. Adding in data like unemployment rates, interest rates, inflation etc....it becomes a huge debacle to determine which variable influences which other variables. All these variables just look like levers to be pulled on this big machine we call the national economy but they're not. They are the expression and result of hundreds of millions of American's making little decisions everyday. People think naively that if I increase Government Spending I will increase employment and in turn increase GDP right? or maybe the effect will be to do the opposite and decrease Consumption in anticipation of future taxes and hence have no effect on GDP. Or for another example of the complexity look again at Government spending and unemployment. Which drives which? Does higher unemployment drive up Government spending or does Government spending drive down unemployment? Which way does the causal relationship run? We don't know but lots of theories abound.
All these doubts and questions as to which direction the causal relationship flows can be more easily answered in microeconomics through experiments. If I wanted to know the relationship between student education levels and family income and wasn't sure which way it ran we could find two groups of people randomly and in one group increase their wages only, the other group increase their education only, (we should also have a third group, a control group) and see what happens. Experimentation is not possible in macroeconomics so all we can do is take historical data, compare countries that took different policies over time, control for differences, etc.....enter Sargent and Sims.
Sargent showed the inseparability between monetary and fiscal policy and the role of expectations and learning. It is a relationship taken for granted now but it was not fully appreciated until a few decades ago. Friedman said that inflation was and always is a monetary phenomena. Yes, but Sargent showed how runaway fiscal policy, i.e. government overspending to get out of a recession, and expectations of future fiscal policy could lead to a massive expansion of the money supply and hence high inflation. In fact, his most famous essay "The Ends of Four Big Inflations" focuses on how Austria, Germany, Poland, and Hungary ended their bouts with hyperinflation after WWI. The graphs are simple and crude but this was before all the uber-high tech computing available today.
Sims is tougher to explain.....I think about Isaac Newton inventing the math he needed to describe the physical world, calculus. At the time there were no mathematical techniques to model movement that was changing in velocity and direction. In a similar way Sims was dissatisfied with the mathematical techniques available 30 years ago and so created Vector Auto Regression (VAR) as an econometric method to deal with time series data where the relationship between the variables is ambiguous. One can also include lead and lag variables to handle the problem such as when does a change in interest rate affect other variables....it could be in the subsequent year but also in years thereafter. Or the announcement of increased Government spending in the future could affect outcomes and behaviors now, this would be a lead variable. The spending next year affects variables this year. So if there were three variables, A, B, and C but we didn't know the exact structure of their relationship we could toss all the historical data for A, B, and C into the VAR machine and it would tell us each of these relationships.
A=B+C (how do B and C affect A)
B=C+A (how do C and A affect B)
C=A+B (how do A and B affect C)
and in how many lag or lead time periods do the variables affect each other. The benefit of this method is that it does not require a ton of prior model specification. The downside is that it always has the feel of data-mining and or of trying every possible combination to get the best fit without regard to spurious relationships. I'm sure a real macro-econometrician could correct me on this.
So all this to say that macro is difficult and complex enough now to plausibly claim that we're still in the Dark Ages. Sargent and Sims have moved us closer to economic Enlightenment. Cheers for their good work!