This is a very long post. It’s the latest version of my thoughts on Keynesian stimulus, the idea that spending creates prosperity or supports our economy or rescues it from the doldrums. John Papola, my collaborator on the video projects at EconStories, continues to push the two of us to think about these issues. So the thoughts that follow have been stimulated greatly by those conversations. Some of my earlier thoughts on stimulus and spending are here , here , and here . This isn’t as well organized as I’d like so read on at your own peril.
In particular, I’ve been thinking about the idea that in times of high unemployment, the government can borrow money and give it to people to spend creating beneficial effects—either job creation or via the idea of a “multiplier,” creating income gains in addition to those received by the people who get the initial gift. And that in turn will create jobs.
For many people, economists and non-economists, nothing could be more intuitive than the idea that giving people money stimulates the economy. I think they have two fundamental ideas in mind, two ideas that are somewhat related. The first is the multiplier. If the government gives A one dollar to spend, A spends it on something B produces, encouraging B to hire more workers. But it doesn’t end there. Now B has money now who spends it on C and so on. Because people save part of what they spend, A’s spending is less than a dollar, say .9. Then B spends .81 cents (.9 x .9) and so on. If you do the math (and if everyone spends 90% of what they earn) then this infinite series leads to a total amount of spending equal to $10.
There are two standard criticisms of this idea. One is that the money has to come from somewhere. If you tax D to raise the dollar, and D has the same spending rate as everyone else of 90 cents for every dollar received, then there is no increase in total spending. If you borrow the money, then some people who are going to pay the taxes will put aside money now in anticipation of future taxes and that will cancel out some or all of the $10 gain.
(I am ignoring the Keynesian implication that savings is just waste. But savings usually funds investments for the future that in turn create output. It’s all spending. But you can argue that in today’s investment climate, extra savings doesn’t fund much investment–people are just sitting on it.)
If people don’t anticipate future taxes or don’t respond to them today, then yes, says the standard argument, you will get $10 of stimulus today, but it will be offset by $10 of losses tomorrow. The Keynesian then responds with the second idea behind using government spending to stimulate the economy. That’s the idea that when we’re in a recession, the economy is broken. The engine of the economy needs a jump-start. The pump needs priming. By giving people money who start spending now, the whole system gets going so it can be once again be healthy and self-sustaining. It’s like the battery of your car. If you drain your battery because you left the lights on overnight, the car needs a jump. Once it gets going, it will recharge simply by driving. So in the economic analogy, once people get spending, the circular flow of spending gets going once again, the people who were unemployed soon get employed, they start spending and the economic engine will once again be healthy and keep going. This is what Keynes meant when he talked about “magneto trouble .”
This metaphor of the car that needs a jump start or the ship that needs a push because it’s hit a patch of water that’s windless, has a certain compelling feel to it. But it really has no intellectual content. It’s an ex-post story to make you feel good about spending money. There’s no there there. So when a Keynesian is confronted with the fact that about $800 billion dollars of spending was promised in February of 2009 (and more than half of it has been spent) but unemployment has barely budged, well that just proves how bad the economy was in February of 2009.
It’s like saying that using these jumper cables will get your car started but sometimes, a regular jump isn’t sufficient. You have to power up the healthy car when you do the jump because the car with the dead battery is so far gone that a regular jump won’t do it. That’s true with cars. But is it true with economies? It might be. It’s possible that the economy was so sick in February of 2009—the cancer had spread more widely than we realized so we needed more and stronger chemo, to pick another metaphor. The problem with the metaphor is that it has no content. With a battery you can measure the strength before you jump it. You can do a CT scan of a cancer patient. But we don’t have a diagnostic tool for the economy other than static measures like the current unemployment rate that tell us nothing about the expected or possible trajectory of the economy over time. So the current Keynesian stories of why the stimulus didn’t work are just stories. Ex-post stories. They can’t be distinguished from the hypothesis that we don’t know what we’re talking about or the hypothesis that stimulus spending is a sham.
I remember a cartoon where the driver of a stalled car gets out, opens the hood and there, underneath the hood, is a giant switch that is set to the “Off” position. So that’s the problem, he says. In fact the economy is nothing like a stalled car. Here we are in the worst recession of my lifetime, the worst since 1933 yet over 90% of the workforce has jobs and millions of jobs are getting created every month. Just not quite enough to match the numbers that are looking for work. The economy isn’t broken, just not working quite as well as it usually does. This is another reason that it’s really hard to know what is wrong and how to fix it. There isn’t an switch we need to turn back to “on.” And it probably isn’t as simple as boosting spending back to some larger level.
Having said all that, the logic of the circular flow idea, that spending begets further spending and that payment to labor leads to demands for products which leads to supply of products is pretty attractive. But part of the circular flow story is misleading because spending–the expenditure and receipt of money—can be very deceptive as a measure of prosperity.
For example, proponents of “buy local” will argue that it’s good because it keeps the spending within the community. It keeps the money within the community. But we don’t care about money, per se. If we only buy from each other (our local neighbors in our small town or even a big city), all the money will stay local but it won’t go very far in both senses of the word. The prices will be high because we won’t be able to fully exploit the division of labor and specialization that comes from trading with a wide group of people. We’ll be very poor if we keep all of our money circulating among ourselves. What we really care about is not money, but what the money can buy. We care about our prosperity, not the dollar value attached to our paycheck. If we only trade among ourselves, our productivity will be relatively low. We will have a low standard of living. We will be poor even though we’ll have all the money. Money by itself, is not a good measure of prosperity.
Money, after all, is just a way to avoid having to barter for stuff and avoid the coincidence of wants. If I want a chicken and potatoes for dinner, I don’t want to find someone who raises chickens and grows potatoes who is also interested in getting a degree in economics. If the person who is interested in a degree in economics is a furniture maker, I can give her an economics lecture in exchange for a chair and then try to find a chicken farmer who needs a chair. Very time-consuming, what economists call high transaction costs. Better to take some general measure of purchasing power, money, from the furniture maker and buy a chicken.
So money is a veil. It hides the underlying reality that what I can consume depends on what I can produce. And what I can produce depends on the people I can exchange with and cooperate with economically. The division of labor is limited by the extent of the market. If I have a lot of people to exchange with, then I can be more specialized and via technology, get a lot richer than if I trade with a small circle of locals. If I trade widely, I’ll have more money, but the amount of money I have is an effect not a cause. The existence of money is a cause–that creates wealth because it allows me to trade without have to find the chicken farmer who wants an economics lecture. But that’s it.
The whole circular flow idea that underpins the Keynesian story is bizarre. It’s a veil that disguises what’s really going on–the trades and exchanges I’m able to make as a worker and consumer. The car analogy is flawed. What keeps the economy going is our mutual creativity and cooperation, not our spending. The spending is just a result of our underlying productivity and desires to consume out of what we produce. Part of the circular flow story is correct–there are a set of forces that keep the whole thing going on its own–but those forces are my skills and productivity and your skills and productivity and how they all fit together–what Arnold Kling has been calling PSST, patterns of sustainable specialization and trade. I would remove the word “sustainable” and stick with PST. The word “sustainable” suggests a transience or fragility that may or may not be important. Suffice it to say for now that in good times, the patterns of specialization and trade are changing constantly, as demand for products changes, as new entrants to the labor force bring different skills to the market, as tastes change and so on.
The web of work and wants that we create as employees, entrepreneurs, managers, and consumers is remarkably sustainable on its own. But it appears to work better in some times than others. In 1997 if you had a computer skill and applied that skill by working in a firm suddenly went out of business through no fault of your own, you could find another job pretty quickly. That was less true in 2001. And even less true now. Losing your job today is much scarier than losing your job in 1997. Having said that, the highly educated friends that I know who lost their jobs in the last three years have found work. It was scary but they didn’t stay unemployed for very long. Longer than they would have liked and one had to move to find a new job, but they both did OK. They weren’t unemployed for 99 weeks.
But there are people who have been unemployed for a very long time, who aren’t being integrated into the existing pattern of specialization and trade. There could be many reasons for that. As I said earlier, this is why it’s so tricky. The labor market isn’t really broken. It’s just not working quite as well as usual.
For now, let’s ignore the reason or reasons that more people than usual are out of work for a lot longer than usual. What I want to consider now is the potential role of various kinds of spending to help those people, what the Keynesians call boosting aggregate demand.
Suppose 5% of the workforce is in the construction business. They’re making houses that people like. Now it turns out people don’t want to buy houses as much as they did before. Forget the reason. Doesn’t matter. A bunch of construction workers are now unemployed. Now consider two scenarios:
1. The unemployed construction workers find something else productive to do. They go and find work, take the money they receive from doing that new thing, and buy shoes.
2. The unemployed construction workers receive checks from the government. They use the money to buy shoes.
What’s the difference? Don’t both scenarios “stimulate the economy?” The difference is that in case 2, we only get more shoes. Maybe. But in case 1, we get more shoes and more of something else. Why is that relevant? Because purchasing power (via cash alone) is not prosperity. We know that via reductio ad absurdum. If the government sends big enough checks to the construction workers (and does so forever, to make it easier to think about) to get them to NOT find a new career, that doesn’t create prosperity. That creates the ability of the construction workers to lay claim to lots of stuff. But they produce nothing. Widen the pool of people who get the checks. That expansion of aggregate demand REDUCES GDP because it takes even more people out of productive activities.
(This is obviously related to Bastiat and the broken window fallacy .)
Or consider this case:
3. The government borrows money and gives it to renters who use the money to buy houses. The government gives away enough money to enough renters so that they buy all the vacant houses and then demand enough new houses to employ all of the out of work construction workers.
I choose this example because a standard criticism of the stimulus is that it was poorly designed. It took too long. Not enough of it was targeted to real production and infrastructure. My question is whether this makes any difference. It’s all supposed to boost aggregate demand and all of it does boost AD in Keynesian terms. Keynes did say and I have heard Joseph Stiglitz say that paying people to dig holes and fill them back in stimulates the economy, it just isn’t as good as doing something productive which stimulates the economy and produces something of value at the same time.
So scenario 3 as I’ve crafted it, is a perfectly designed stimulus plan–it’s targeted and puts the people back to work, right?
It is worth mentioning that while Keynesians like to talk about sticky wages, nobody thinks housing prices are terribly sticky. They’re a little sticky. Someone who paid $1 million for a house and who is moving and can’t find a buyer to pay the $1.2 million he was expecting to collect will eventually lower the price. It may take a while because the seller isn’t sure of the market price. But he will eventually catch on and lower his price. Look at the data. The price of houses is falling. And that’s in a world where the government is artificially propping up the price as best as it can via Fannie Mae, Freddie Mac, the FHA, and artificially low interest rates.
So if the government would let those prices fall, that would clear out the stock of empty houses fairly quickly and help get those construction workers back to work without any Scenario #3 stimulus at all.
What’s the difference between these two cases? Are they the same with respect to construction employment?
Here’s the problem with case #3. Suppose because of really bad housing policy and bad financial policy, there is a huge expansion of construction employment. When the housing market collapses, suddenly there are a lot of people who are unemployed. We can put them back to work by paying renters enough money to buy up the stock of empty houses and to build enough new houses to soak up all those unemployed workers. That would put the workers back to work but at a very high cost. But we don’t want to re-create that world. That’s the fundamental problem–too many people were allocated to housing (and too much lumber and too many bricks and so on). To sustain that pattern of specialization is trade is very expensive–it means continuing to over-produce houses to keep those workers employed.
If the government would let the price of housing fall then it would also eventually help some of the unemployed construction workers get back to work as the stock of empty houses disappeared and eventually a demand for new houses would return. But the price wouldn’t fall low enough to put the whole mass of unemployed construction workers back to work building houses.
The deeper puzzle is this. In good times, industries are rising and falling, jobs are being created and destroyed. But in good times, more jobs are being created than destroyed. In bad times, the opposite is the case. What is going on today that makes it hard for workers to find new opportunities and what happens in good times to make it easy? Why is it so hard for construction workers and others who are unemployed to find ways to use their skills? It wasn’t hard in 1999 when millions of jobs were also being destroyed.
Ironically, I think the fundamental reason that Keynesian stimulus is mostly ineffective is the anxiety decision-makers have about the future. In good times, when someone buys a pair of shoes, the shoemaker is encouraged and considers making more shoes. The shoemaker considers hiring more workers or expanding or buying better shoe-making equipment. But in bad times, those actions which are risky in the best of times look even riskier. So the circular flow mechanism of spending begetting more spending, doesn’t work in quite the same way. And reason is that the underlying patterns of specialization and trade get harder to re-establish in bad times. And that is true because the people with capital, the people who make decisions about expansion get more risk-averse about the future.
So the fundamental stimulus question is this: does government borrowing a lot of money and spending it on tax rebates and grants to the states and some infrastructure (not much) and some other things, does that make people less confident or more confident about the future. Who knows? There’s no easy way to measure confidence, especially among the people who are relevant, those who are considering hiring new workers. But it is plausible that running up much bigger deficits reduces confidence and increases anxiety. It is plausible that the whole Keynesian circular flow story breaks down or at least works less effectively when people are anxious about the future.
The bottom line for me is that Keynes was very right about one thing–people’s perception of the future plays a big role in their willingness to take risks. What he was wrong about is the ability to use government spending as a way to encourage a positive perception of the future or to produce real economic results that lead to prosperity and increased employment.