Wednesday, November 2, 2011

Keynesian Predictions about Saving

Tyler Cowen writes:

"I will ask the Old Keynesians, here is the report from last week:

“The lack of income growth meant that the boost in spending came from a decline in the savings rate,” said David Semmens, US economist at Standard Chartered.

…The savings rate fell 0.5 per cent in the period, reaching a level last seen at the end of 2007.

I will predict that the savings rate does not substantially rise next quarter. Absolute savings will go up over some time horizon but I don’t expect it to do so immediately for cyclical reasons. Are the Old Keynesians now expecting savings to go up in some manner, due to “the paradox of savings”? After all, spending has risen. Here is a chance for Old Keynesian economics to score some points, so what will happen with savings? What does a liquidity trap model predict? Should savings be higher already, or does it take longer for that effect to kick in? How long?

Note that the New Keynesians generally do not have much to do with the paradox of savings; my view is that it may have applied to parts of 2008-2009 but is unlikely to apply today.

When it comes to Old Keynesian predictions I don’t want to count “money matters” (formerly an anti-Keynesian prediction by the way),”the recession was really bad,” or beating up on various odd right-wing views. I want forcing predictions which discriminate against the best available alternative theories.

In the comments I also would like to hear Keynesian accounts, New or Old, of why gdp growth was suddenly 2.5%, not a great number for a recovery in absolute terms, but was it the predicted number or direction, given that the stimulus was finally exhausted? I am comfortable citing “noise,” but does a liquidity trap model offer this same freedom? Doesn’t the model itself imply that one margin is what matters and you are truly “trapped”?

Drawing really long time-lines to obscure false predictions of the moment does not count as an answer!"


I'm not sure why he questions the freedom that the liquidity trap model offers with respect to answering "noise". Any model is an approximation in the first place, and even if it were exactly right, curves shift anyway. So I'm very comfortable citing "noise", by which I just mean "I don't have a crystal ball that allows me to predict what happens from quarter to quarter, and I think it's more important to look at trends". I don't think this offers any immediate reason to think that savings will go up as a result of the lower savings rate because of the paradox-of- thrift-in-reverse. The liquidity trap we're facing at this point in time is a zero-lower-bound trap (no, the liquidity trap does not have to occur at the zero lower bound). So let's think of a simple loanable funds model presentation of that:



If the supply of loanable funds shifts to the left, from S to S', it obviously doesn't really change the level of investment as long as we're at the zero lower bound. That's what I meant when I've spoken previously of an interest rate that is "too high" because of liquidity preference as a price floor. Think of another price floor like the minimum wage. If some unemployed youth decide to exit the labor market when there's a binding minimum wage, do we expect a large change in employment? As long as the minimum wage is still binding, no. And this price floor in the loanable funds market is still binding as well. Holding cash earns a better return (zero) than the market equilibrium interest rate, so the interest rate doesn't fall below zero.

This is just a couple lines I threw together to help think about the liquidity trap, though. It doesn't tell us much about what to expect next quarter. Maybe if people are consuming more it'll be enough to increase incomes and increase total savings despite the lower savings rate. I don't know. But our real source of demand deficiency is of course weak investment. Now, maybe investment demand will pick up if people are consuming more. But you invest if you expect people to consume more in the future. Does the current blip in consumption suggest an increase in future consumption? I don't know that - but knowing that is crucial to answering Tyler's question. I have a hard time making too much of a decline in savings when the real problem we have with demand is an investment demand problem. Moving the S curve to the left doesn't seem as beneficial to me as moving the I curve to the right. Maybe if you cut your savings far enough the loanable funds market will actually clear! But that's a pretty perverse market clearance - it's still going to clear at the point where the I curve hits the x-axis. That's not where we want the market to clear.

So my answer is "noise". I can see where Tyler is seeing some potential. Maybe this bodes well for consumption which may say something about future consumption, which may say something for investment demand. But I don't have a crystal ball and I've never claimed to. What I have are a few simple tools for thinking about the world and past experience and data to put a little more meat on them.

3 comments:

  1. One thing I hate about bloggers is when they start going off on one data point. I feel like Cowen does that a lot. It's also ridiculous that he pays lip service to the possibility of "noise" but then insists that discussing trends is not valid.

    Anyway, not sure what the point about the paradox of thrift is. This is an open economy with foreigners significantly (or perhaps totally) offsetting fluctuations in private savings rates and that adds a dynamic I don't think a paradox of thrift takes into account.

    ReplyDelete
  2. This is an open economic system with and also the considerably (or perhaps totally) offsetting variations in private benefits rates and that gives a energetic I don't think a paraseo of music considers.

    ReplyDelete
  3. I feel like Cowen does that a lot. It's also ridiculous that he pays lip service to the possibility of "noise" but then insists that discussing trends is not valid.saç ekimi öncesi ve sonrasi

    ReplyDelete

All anonymous comments will be deleted. Consistent pseudonyms are fine.