Marcus Nunes directed me to a post where Olivier Blanchard recommends some changes in the way we teach macro. In my view, all of these changes are in exactly the wrong direction, although I don't doubt that far more economists agree with the highly distinguished Blanchard than with me.
The IS relation remains the key to understanding short-run movements in output. In the short run, the demand for goods determines the level of output. A desire by people to save more leads to a decrease in demand and, in turn, a decrease in output. Except in exceptional circumstances, the same is true of fiscal consolidation.
In my 35 years of teaching macro, I never covered the IS/LM model. Instead, I used the AS/AD model (which Blanchard wants to eliminate.) The primary flaw in the IS relationship is that economists often misuse it, and engage in "reasoning from a price change." That is, they typically assume that changes in the interest rate reflect changes in Fed policy, with lower rates implying easier money. This is false. For instance, interest rates fell in late 2007 and early 2008 because of a weakening economy, not easy money. Admittedly the IS relationship can accommodate this possibility through a shift in the IS curve, but it's clear that most economists don't see this, as they frequently refer to low interest rates as easy money.
It is also false that an increased desire to save leads to lower demand and lower output. As Nick Rowe tirelessly points out, it is an increase in hoarding that matters, and whether that increase is accommodated by the central bank. Of course if one views the interest rate as "monetary policy" then it's easy to see why people could fall into this trap. When there is an increased propensity to save, then the central bank must reduce the money supply in order to hit its interest rate target. In fact, central banks target inflation, in which case an increased desire to save has no impact on demand.
The statement about fiscal consolidation is also wrong. Or at best you could say that fiscal consolidation is not contractionary, except at the zero bound. In other words, rather than the model only failing in "exceptional circumstances", Blanchard's model only holds true in exceptional circumstances. And this isn't just my view, as recently as 5 years ago Paul Krugman held the same view as I do.
The demand for goods, in turn, depends on the rate at which people and firms can borrow (not the policy rate set by the central bank, more on this below) and on expectations of the future. John Maynard Keynes rightly insisted on the role of animal spirits. Uncertainty, pessimism, justified or not, decrease demand and can be largely self-fulfilling. Worries about future prospects feed back to decisions today. Such worries are probably the source of our slow recovery.
This has things exactly backwards. Monetary policy determines the rate of growth in NGDP, which in turn determines the level of nominal interest rates. Interest rates are (for the most part) not the cause; they are the effect, an epiphenomenon.
The LM relation, in its traditional formulation, is the relic of a time when central banks focused on the money supply rather than the interest rate. In that formulation, an increase in output leads to an increase in the demand for money and a mechanical increase in the interest rate. The reality is now different. Central banks think of the policy rate as their main instrument and adjust the money supply to achieve it. Thus, the LM equation must be replaced, quite simply, by the choice of the policy rate by the central bank, subject to the zero lower bound.
In my view this is exactly the opposite of the lesson that we have recently learned. It's clear that zero rates will occur in future recessions. Thus we need to stop using a policy instrument that freezes up every time we need it, and replace it with a policy instrument that is not subject to the zero bound problem. I suggest using the price of NGDP futures, but there are other possibilities. For instance, Singapore uses the exchange rate. If the US is too big to use the exchange rate, there are always possibilities such as a basket of commodity prices, or some very broad measure of money. But please, anything but a policy instrument that is susceptible to the zero bound problem, which has been the biggest problem with central banking over the past decade. It's no coincidence that the one developed country that avoided the zero bound (Australia) also avoided the 2008-09 recession.
Furthermore, if we replace the LM equation with an interest rate, then students will be even more inclined to reason from a price change. No longer will a leftward shift in the IS curve lead to lower interest rates. Thus all declines in interest rates will look like easy money policies.
If anything, the crisis has shown the importance of the financial system for macroeconomics.
No, it's shown the importance of keeping NGDP expectations growing along a slow but stable path. Financial distress is mostly (albeit not entirely) an effect of NGDP instability.
Turning to the supply side, the contraption known as the aggregate demand-aggregate supply model should be eliminated. It is clunky and, for good reasons, undergraduates find it difficult to understand. Its main point is to show how output naturally returns to potential with no change in policy, through a mechanism that appears marginally relevant in practice: Lower output leads to a lower price level, which leads, for a given money stock, to a higher real money stock, which leads to a lower interest rate, which leads to higher demand and higher output. This is a long, convoluted chain of events with doubtful realism. Central to the adjustment is the assumption of constancy of the nominal money supply, which again is not the way central banks do business. And the notion that economies naturally return to normal has not held up well over the last seven years.
Unemployment in America is 4.7%, whereas the average rate over the past 50 years is 6.2%. Who says economies can't return to normal? I would do the opposite, eliminate the IS/LM model, and rely solely on AS/AD. I do agree that the interest rate-oriented transmission mechanism used by Keynesians is a confusing mess, and inaccurate. The much simpler monetarist version is far superior. The AD curve is simply a rectangular hyperbola, a given amount of NGDP, determined by the central bank. Nominal wages are sticky. When NGDP changes unexpectedly, there is a change in hours worked, as wages adjust slowly to reflect the new level of NGDP. Once wages have fully adjusted, output returns to the natural rate. This model has done an excellent job of explaining the economy since 2008.
Unfortunately, Blanchard wants to replace the AS/AD model with a much inferior Phillips Curve model, which focuses on demand shocks and makes it harder to see supply shocks. And even worse, this is not the Friedman version where unexpected inflation affects output, but a much inferior Keynesian version where low output reduces inflation:
These difficulties are avoided if one simply uses a Phillips Curve (PC) relation to characterize the supply side. Potential output, or equivalently, the natural rate of unemployment, is determined by the interaction between wage setting and price setting. Output above potential, or unemployment below the natural rate, puts upward pressure on inflation. The nature of the pressure depends on the formation of expectations, an issue central to current developments. If people expect inflation to be the same as in the recent past, pressure takes the form of an increase in the inflation rate. If people expect inflation to be roughly constant as seems to be the case today, then pressure takes the form of higher--rather than increasing--inflation. What happens to the economy, whether it returns to its historical trend, then depends on how the central bank adjusts the policy rate in response to this inflation pressure.
If we've learned anything recently, it's that low output does not lead to ever falling inflation. The Keynesians have reversed causation. NGDP shocks impact both inflation and output. This "musical chairs" interpretation of monetary non-neutrality has done vastly better than the Keynesian version during the Great Recession.
In my view macro needs to head in a different direction. We need to remove nominal interest rates and inflation from their central role in our models. We also need to de-emphasize the role of the financial system, which currently plays far too large a role in our money and banking textbooks. Instead we need to focus on a model with these characteristics:
1. A simple AS/AD model, where output can change due to supply or demand shocks.
2. AD is equal to NGDP and determined by monetary policy, which is implemented through changes in the supply of base money, not interest rates. NGDP futures prices are the best policy instrument, and also the best indicator of whether money is easy or tight.
3. Nominal hourly wages are sticky in the short run, leading to short run monetary non-neutrality, and flexible in the long run, leading to a self-correcting mechanism. If you want to add a bit of hysteresis that's fine, but I believe its importance is overrated.
In my view, if macroeconomists had adopted this framework in 2007, the Great Recession never would have happened (but the Great Stagnation would still have happened.)
PS. Over at TheMoneyIllusion, I use graphs to explain some of the ideas in this post.