[In this third lecture of his American Economic Policy class at Harvard given on January 29th, Martin Feldstein explains how he and the profession turned away from Keynesian economics, and how he made his way back.]
This morning the Department of Commerce published preliminary estimates of what happened to GDP in the fourth quarter of last year. These estimates will be repeatedly revised, “but you’re a policy maker and you know only what you know, and this is what you know.”
The estimates are seasonally adjusted and expressed as annual rate after controlling for price changes. Growth was pretty crummy at an annual rate of 0.7%, down from an annual rate of 2.0% during the previous quarter. Should we be worried? Are we slipping into a new recession?
Consumer spending was growing at 2.2%, contributing 1.5% to GDP growth, and residential construction was good. But fixed investment in structures and equipment was negative and pulled growth down, as did exports and changes in inventories. These negatives added up to 1.2, the pluses to 1.8, so we are left with about 0.7% for overall growth.
This gives you a picture of what was going on. The folks at the Fed have to ask if they will keep raising interest rates.
This is the great thing about being in economics: there is always new stuff happening and new information being released to help you figure it out.
Keynesian economics as a turning point
Keynesian economics changed the way in which policy makers and the public thought about making economic policy.
The publication of John Maynard Keynes’s General Theory of Employment, Interest and Money in 1936 was a turning point for macro-economic policy.
In this lecture we will see why macro policy came to be seen as the solution to business cycle problems, why economists lost faith, and why it came back temporarily.
Before Keynes there was no theory of fiscal policy, balanced budgets were seen as normal. Besides government spending was not that big, amounting to only 3% of GDP. During the depression there were public works projects of all sorts to hire people and give them jobs, be they construction workers, artists, musicians, bridge builders and artists alike. But there was no strategy that this would lead to changes in the gap between actual GDP and its potential, the idea of the multiplier was not there.
This idea came not from Keynes, but from Richard Kahn who later became Lord Kahn.
What about monetary policy? Before Keynes it focused on keeping inflation low, avoiding financial crises, and protecting the stock of gold.
There was no sense of counter-cyclical policy. To the extent that there was, it was focused on gold. At that time paper currency was backed by gold. If the government was losing gold because people feared higher inflation and a debasement of the currency then to maintain its stocks of gold it would raise interest rates, dampen the economy and inflation.
That was it for macro policy. The public didn’t expect more, not thinking the government had the capacity to do anything. The mindset was the same as we would think about the weather, we don’t expect the government to do anything about the weather.
Government policy during a recession was meant to be relief through public work projects. There was no idea of “fiscal policy.”
What John Maynard Keynes said is that unemployment was caused by a lack of aggregate demand. Well don’t we all know that? Some say unemployment is caused by people not wanting to work, or wages being too high, and that it will sort itself out. Wages will fall, prices fall, and real monetary stock will rise. Arthur Pigou was a proponent of this way of thinking about things in Keynes’s time, and Keynes took aim at him, saying we just can’t keep waiting: “in the long run we are all dead.”
The basic idea Keynes brought to policy makers was the “multiplier.” This was really the work of Richard Kahn who pointed out [in a 1931 paper published in the Economic Journal] that government spending increases GDP, and this rise in income leads to more consumer spending, which in turn raises income and kicks in another round of increased consumer spending. This process will keep going, though at a reduced rate in each successive stage.
Higher permanent government spending continually starts a new stream of income-consumption, income-consumption cycles. Eventually the process will settle down, and there will be a change in real GDP equal to the change in government spending divided by 1 minus the marginal propensity to consume [the fraction of an extra dollar of income that is spent].
This is the beautiful multiplier. If the marginal propensity to consume is 0.8 [that is if consumers spend 80 cents of an extra dollar of income, saving only 20 cents] then a permanent increase in government spending causes GDP to go up five-fold. Wow, think about that!
A great idea, but then the evidence came along, which in the US case was the military build up in preparation for WWII. Throughout the 1930s the unemployment rate remained high, until WWII. In 1939 the military buildup began, with 1.5% of GDP or $1.5 billion in military spending. In 1941 this increased to 5% of GDP, in 1943 we were all in and 32% of GDP was in military spending. Amazingly non military GDP was 14% greater than in 1939. And total GDP went up because of both defense and non defense spending, showing the power of the multiplier.
Counter cyclical fiscal policy
Economists came to appreciate, came to respect, this new theory. After WWII American Keynesians entered Harvard. Alvin Hansen was the great disciple, and he took things a step further and taught that we can use Keynesian policy to eliminate the business cycle.
All we have to do is calculate the gap between GDP and its potential, and use fiscal policy to fill it.
Lets be more specific. If at full employment output the unemployment rate is about 5%. Full employment doesn’t mean that everyone is working. Some people are between jobs and looking, so there is unemployment because of job search. And there are also new entrants, quitters, and those temporarily laid-off. But if we see an actual unemployment rate of 8%, then the excess unemployment rate is 3%. But how do we get from this to a gap in GDP?
We are helped by Arthur Okun, a former chairman and member of the Council of Economic Advisors during the 1960s. He wrote a paper [called “Potential GNP: its measurement and significance“] that showed every 1% cyclical increase or decrease in rate of unemployment corresponds to 2.25% change in real GDP. This got the label Okun’s Law. There is no reason it should hold, but that seems to be the case. Okun’s Law is really a rule of thumb. “I wish there was a Feldstein’s law, but I haven’t done anything to merit a law.”
So a gap of 3% in the unemployment rate leads to a GDP gap of 6.75%. If potential GDP is $10 trillion, then the gap is $675 billion, and we need 675/5 or $135 billion of spending. [I’m not certain I copied this correctly off the board.]
Aren’t we clever. Then we can go even further and predict the future path of GDP to see what happens if there is no change in policy, adjust policy, and get even more income. Wow. Sounds good. Sounds good.
This is the hope economists had in the 1960s. If we build better economic forecasting model, learn the multipliers (including dynamic multipliers), then we will be able to get rid of this terrible thing that hurt the economy and people in the 1930s.
Retreat from Keynesian economics
Somehow within a decade we had lost faith in all this. Fine tuning of the economy just wasn’t working.
Why did that happen? In the 1960s everyone thought in this way. President Nixon even went so far as to say that we were all Keynesians now. So what happened?
First, the evidence turned against this way of thinking. In the late 1960s the unemployment rate was 4%, and the inflation rate was 5%. In the late 1970s after applying Keynesian economics, the unemployment rate increased to 7% and inflation was 9%. So it was a failure. It wasn’t just the public that judged this poorly, but also the profession which concluded that fiscal policy wasn’t all it was cracked up to be. Why?
First, the fiscal multiplier was much smaller than textbook multiplier imagined it, not at all 5 as in my first textbook. By the 1970s econometric studies suggest it was just 1. Why was the multiplier so small? The typical policy changes, for example in taxes, were not permanent, they were temporary. One or two-year tax cuts lead to consumer responses based on what Milton Friedman called “permanent income.” Experience shows that people base their spending decisions upon some sense of their longer run income, and will not spend the entire increase in income that is temporary. Only a fraction of lottery winnings or one-off tax changes is spent, the rest is saved. Not everyone responds this way. The unemployed and low-income people may not, but they are not tax payers so not likely to be affected by tax changes anyways.
What about government spending? Most government purchases are defense related, the rest of government spending is transfer payments. The thing we have evidence on is defense spending. There is a multiplier to this spending, but not transfer payments.
Second, even the direction of the impact of changes in fiscal deficits was uncertain. Why?
Lets’ say government increases spending or cuts taxes, there is an immediate positive effect. But this also influences market expectations of the future: maybe the deficit in the future will be larger, so maybe long-term interest rates will be higher, and this will dampen aggregate demand now. An increase in government spending today leads the market to believe larger deficits in the future. But also expectations of even higher taxes in the future to deal with those larger future deficits. People thinking in terms of their permanent income say that the short-term effect is offset by longer run changes.
Third, and most important, is the lags in the fiscal policy process. What are they?
- A recognition lag. It takes time to recognize that we are in a downturn as we don’t know immediately if changes in GDP are temporary or permanent. We need time to figure this out. For example in 2007 the Fed’s minutes show that it took them a year to appreciate the magnitude of the change, and that this was a serious downturn.
- An implementation lag. The Fed at least can operate any time it wants, but fiscal policy requires agreements in the Congress and between Congress in The White House. So the legislation can lag significantly. Change don’t happen overnight. Tax changes have to be implemented, and spending changes also require time to design programs. Even after this people don’t rush out and start spending any new income immediately.
So if we look at recent experience we see how all this played out in practice. Housing prices turned down in 2006, the economy peaked in December 2007, and it took months to recognize this as a peak. A small stimulus was introduced in 2008, and a bigger one in 2009, fully three years after the peak. It was promoted as shovel ready projects, but that was not true. In 2009 The White House and Congress passed an $800 billion package over three years: 200, then 400, and then another 200 billion. But this took a long delay in getting going, only in 2010 did they spend $400 billion.
Why is this important? Think of the business cycle. If the peak to trough is on average 10 months, then stimulus tends to come along when the economy has already turned the corner, and actually adds instability to the process.
One of my favourite papers is by Milton Friedman that looks at lags in policy and the impacts on the economy. It is not even clear that fiscal policy will be working in the right direction. Every time there is debate, every time there is uncertainly, the argument is put forward that policy makers have to give it a try, and at least do something. At least we are moving in the right direction, the argument continues. Friedman shows this attitude can even be destabilizing, and make things worse. If you are going to use a very active policy relative to the size of the problem, then you must have very good aim. If your timing is not good, then you are better off not doing anything.
This is likely to be the case if you face a short downturn. Your aim is not going to be very good. But in 2009 the recession was very large and we could use discretionary policy and expect to do well. The recession was so big, so deep, so long-lasting, that we can set aside our usual hesitations and use fiscal policy.
In late 2007 I realized that this was a different game, and fiscal policy was needed. This was a disturbing conclusion. I had to change my mind. It was a disturbing thing to move away from relying on monetary policy. But I changed my mind.
So how to do it? The best way to do it quickly is not through more government spending, but through a personal tax cut. Basically in advocating this and agreeing with the Bush administration I was disregarding Friedman’s lessons on the permanent income hypothesis. It turned out I was wrong, and the impact of the temporary tax cut was small. Larry Summers also concluded this.
The size of the downturn amounted to a gap in GDP of $700 billion. But what we got in response was a Bush tax cut in 2008, a rebate. I was an a cheerleader for it, inappropriately as it turned out. This amounted to a stimulus of $78 billion and a $12 billion increase in consumer spending … basically nothing.
There we were coming into 2009 and the Obama administration recognized that we needed fiscal policy, and in the next lecture I will review that experience, and then shift to long run policy.
[Heard among the big football types sitting in front of me at the end of class: “it is pretty cool that he can disagree with himself, admit that he was wrong”, “Yeah, Mankiw would never admit he was wrong.”]