The recent financial crisis and great recession has spurred an ongoing economics blog war. The blog Noahpinion keeps a running summary. The battle is largely between the Keynesian view of macroeconomics, spearheaded by Paul Krugman, versus the Chicago school, represented by the likes of John Cochrane. In very simple terms, the Keynesian explanation of recessions are that they are due to decreases in aggregate demand. By that, it simply means that economic activity shrinks due to some shock like the bursting of the real estate bubble or an inexplicable decrease in consumer confidence (Keynes called them animal spirits). Keynesian’s think in terms of reduced phenomenological models with names like the IS/LM (Investment-savings/Liquidity preference-Money supply) model or the AD/AS (aggregate demand/ aggregate supply) model. These models apply supply and demand ideas of microeconomics to the entire economy. Although the models are simple, they can give very specific prescriptions for what to do about certain economic situations. The AD/AS model is analogous to demand/supply curves of microeconomics for single products. It consists of an AD curve and an AS curve on a graph of global price level versus global output (i.e. GDP). The AD curve slopes down to the right since the higher the price the lower the demand for goods and services while the AS curve slopes up to the right since the higher the aggregate price the higher the aggregate supply.
Unlike in microeconomics, the reasons for why these curves do what they do is not completely self-evident and takes some reasoning to justify. The problem in a recession is that the entire demand curve shifts to the left so the equilibrium GDP also shifts to a lower level. Keynesians argue that in order to get out of a recession, we need to move the demand curve back to the right and the simplest way to do that is to increase demand through government spending. This was the rationale for the 2009 stimulus although many of the Keynesians like Paul Krugman and Christina Romer (who was in the Obama administration) cautioned beforehand that the amount was too small to get us completely out of the recession. Hence, the fact that we are still in recession is not evidence that the stimulus failed.
The Chicago school doesn’t believe in simple reduced models. They prefer to build macroeconomic models from microeconomic models. This requires knowing how individuals will behave which is very difficult if not impossible. To get around this problem, they believe in the efficient market hypothesis. This hypothesis is very seductive because it allows them to use the tools of stochastic analysis and fixed point theorems to prove things. However, if the market is efficient then there cannot be sudden irrational shifts in aggregate demand. The recession can’t be caused by the bursting of a bubble because that would indicate a market failure. Hence, the only reason we have recessions is because of structural problems, such as bad regulations or taxes that lead to the misallocation of resources. For them, the only way out is to remove these regulations and let the economy heal itself. They believe that government spending will only crowd out private spending. Prior to the great recession, the Keynesian view was the standard one for macroeconomics but recently policy makers, especially in Europe, have gone against Keynesian ideas. Europe embraced austerity as a means of instilling private sector confidence as a cure for the recession. However, this has largely failed and the backlash against austerity is beginning just as the backlash against stimulus was waged in the US.
I want to point out two things with regard to this debate. The first is that aggregate demand is a marginal attractor and hence it is not unreasonable that small perturbations could move it. Here, I use “marginal” in the physics sense and not the economics sense to mean a neutrally stable attractor. In other words, the aggregate demand curve could exist in an infinite number of locations. The reason is quite simple and that is the economy scales with the population. If we removed half the population, we would reduce our economy by about half provided that we removed people uniformly. This invariance is not perfect of course because there are some things that don’t scale exactly but it probably is true for small changes. It’s also easy to see how a reduction in one sector of the economy could lead to a shift in the entire economy. If all the construction workers were suddenly laid off then they would spend less, which would mean less demand for food trucks, and lumber and so forth. The new equilibrium could then be an economy with a smaller number of people participating. If one continues the argument, it is also not difficult to see that injecting money into the economy could possibly nudge it back to a higher equilibrium. In fact, if you really believe in too little aggregate demand, it wouldn’t even matter where you injected the money. Random stimuli may even work better.
The second thing is that there is a similar top-down versus bottom-up dichotomy in how to model biological systems although it is not as clear-cut and people sometimes play on both teams. One side believes in building highly detailed microscopic models to illuminate the macroscopic behavior. They believe that if you just measure enough things and build a detailed enough model, you could predict diseases or understand the immune system The other side believes in reduced phenomenological models that try to address some specific question. This is mostly where I sit. My work on steroid-mediated gene expression (e.g. see here) is an example of how knowing all the individual pathways would not tell us why the dose-response curve of the gene product obeys a first order Hill function. Kevin Hall and I have shown that you can make very good long-term predictions of body weight change using very simple two-dimensional models based on a few well measured global parameters (e.g. see here). In this era of big data, the bottom-up side is winning the battle for attention and resources. However, I predict that reduced models will make a strong comeback.