How did "fiscal policy" get to be the answer to a question about time and money?
Sophisticated Keynesians will object if I treat "Keynesian" as synonymous with "having a predilection for fiscal policy". But I think they would recognise that the two are correlated. And that correlation is puzzling. The correlation seemed to disappear during the "New Keynesian" era, as the New Keynesians chose monetary policy over fiscal policy; but it reappeared with the recent recession.
Assume a closed economy, for simplicity.
Y=C+I+G is the way we usually do the accounting. But there are many different ways to do the accounting, and each way frames the question differently. Here's a better way:
C means consumption and I means investment. The subscript p means private and g means government.
And here are two different ways we could aggregate terms:
1. Y= (Cp+Cg) + (Ip+Ig) = consumption spending + investment spending
2. Y= (Cp+Ip) + (Cg+Ig) = private spending + government spending
And there are two different questions we could ask about the composition of output:
1. What is the optimal mix between consumption and investment spending?
2. What is the optimal mix between private and government spending?
Those two questions are orthogonal. They are two different dimensions. Consumption/investment is North/South; private/government is East/West. A change in the optimal mix between consumption and investment spending has no obvious implications in either direction for the optimal mix between private and government spending. (Maybe the government has a comparative advantage in investment spending and the private sector a comparative advantage in consumption spending, or maybe it's the other way around.)
Keynes' question was about the first of those two questions, and how a decentralised economy might not get the answer right, and could cause even worse problems as a side-effect of not getting the answer right.
It would be very surprising if getting the right answer to the first question required us to change the answer to the second question. But that is what the "Keynesians" seem to be telling us.
Suppose the economy initially had the right mix on both dimensions.
Then suddenly people decided they wanted fewer new cars and more new buses. Assume cars are bought privately and buses are bought by the government. People cannot signal that change in preferences through the market. They can only signal it at the ballot box, or by talking to government. But we know what the answer is: a reduction in private spending and an increase in government spending. Suppose the government failed to give the right answer, and didn't increase spending on buses. Would that plunge the economy into recession? I don't think it would. People wouldn't be as happy driving cars as they could have been riding buses, but they wouldn't all be unemployed. That wasn't Keynes' question.
Now suppose people people decided they wanted fewer new cars today and more new cars next year. We know what ought to happen. Car producers should reduce current production of cars, and divert resources into investment so they can produce more cars next year. But will that actually happen? Well, maybe if people placed their orders for next year's cars now, that is exactly what would happen. Car producers would see increased orders for next year, and know they should divert resources away from current production towards future production. But if people don't actually give that signal, how will car producers know what they want? All the car producers know is that demand for this year's cars has fallen. If they think this fall in demand will persist, they will actually reduce investment, not increase it as they should. And that might cause a recession. That was Keynes' question. [Update: see Daniel Kuehn's post where he cites a long passage from the General Theory to back up my claim that this was Keynes' question. I had only vaguely remembered that passage.]
"Let's have the government buy more buses so the car producers switch to producing buses instead" isn't the right answer to that question.
But why did Keynes think the market economy might not be able to answer his question correctly? Where's the coordination failure?
I think there are two reasons: a micro reason; and a macro/money reason.
The micro reason: even if producers knew that people wanted to buy fewer cars this year and more goods next year, they might not know whether people plan to buy more cars next year or more bikes next year. And it might not be next year that they want more cars; it might be the year after next. It would take a good forecast based on a market survey to figure that one out.
The macro/money reason: If people had a choice between spending their income on newly-produced consumer goods and newly-produced investment goods, and nothing else, there would be no recession. Even if we add newly-issued stocks and bonds into the list of things that people can spend their income on, it makes no difference, because the firms that issued those stocks and bonds would be buying the newly-produced investment goods. Even if we add previously issued stocks and bonds, and land, and previously produced goods to the list, it makes no difference. If one person is buying used furniture, another must be selling used furniture. Say's Law almost works.
It's when we add "money" to the list of things people can spend their income on, that Say's Law fails. Because if your income comes to you in the form of money, nobody can stop you spending part of your income on money -- you just don't spend it. And if everyone does the same, then we get a recession.
The micro reason, and the macro/money reason, are related. Money is the most liquid of all goods. We don't want to place an order this year for a new car next year. We want to keep our options open, which means we want liquidity. We don't want to decide now what make of car to buy, especially since we don't even know what will be on the market next year. Or maybe to buy it the year after next. Or maybe to buy a bike instead.
It is very hard to coordinate plans when everyone is waiting for everyone else to make up their minds. We all want to keep our options open, but liquidity has a social cost. Perhaps the answer to Keynes' question is that liquidity is underpriced?
The only defence for giving a "Keynesian" answer to Keynes' question is that you don't know the right answer, and your answer is better than nothing.(A variation on a Leijonhufvudian theme.)