Dynamic Inconsistency and Money Demand

Robert Waldmann is trying to write coherent English

I might be able to explain the topic of the paper on which I want comments. Here’s my current effort.

People often claim to have trouble controlling their spending. Thus as they plan future spending and saving, they would like to restrain their future selves. This can be interpreted in terms of utility functions (anything can). Such a view makes sense if we maximize the discounted stream of future pleasure minus pain *and* the rate which we use to discount the future depends on how far in the future it is. This is true if the difference between right now and in one month feels much bigger than the difference between twelve months from now and 13 months from now.

If this is the way I feel about the future I will want to protect Robert in 13 months) from the impatience of Robert in 12 months — that is I will want to prevent Robert in 12 months from consuming as much as he would choose to consume.

Now money is by definition, an asset which we can use to buy something right now (that is it is liquid). To the extent that we are trying to restrain ourselves from spending too much, we have lower preference for mone. We might even prefer a non liquid asset just because we can’t spend it until it matures. We might prefer a large house to cash in the bank, because the large house will last and the cash might be spent by our near future selves leaving our distant future selves without cash or a house.

Now there just happens to be this wonderful survey in which people were asked how much they demanded to get 1000 guilders in 3 months not right then and how much they demanded to get 1000 guilders in 12 months not right then. Thus there is a 3 month rate of impatience and a 12 month rate of impatience and a term structure. The 3 month rate of impatience can be annualized just as a 3 month rate of interest (annual i = (1+3 month i)^4 – 1). If the annualized 3 month rate of impatience is greater than the 12 month rate of impatience, then the agents will benefit if they can find a way to restrict their spending. They want to protect themselves in 12 months from themselves in 3, 6 and 9 months and presumably all period from now to 12 mnths from now. They can do this by keeping their wealth illiquid so it can’t be spent immediately. That is by reducing their money holdings.

On average the annualised rate of impatience corresponding to the 3 month rate of impatience is greater than the 12 month rate of impatience (that is agents make claims such that they should have self control problems). This is not true of all people and, when true, is more true of some than of others.

Our hypothsis is that agents who report impatience rates such that they should have worse self control problems should hold a smaller fraction of their wealth as money. The null hypothis is that answers on impatience have nothing to do with the demand for money given wealth and income and stuff.

The null hypothesis is rejected in favour of our hypothesis.

update: Below find the abstract on the web except for a typor corrected thanks to Travis in comments.

From the archives (please comment)

Dynamically Inconsistent Preferences and Money Demand
Millemaci Emanuele and Waldmann Robert J.,
CEIS Research Paper, 129, September 2008

This paper focuses on two main issues. First, we find that, on average, households’ discount rates decline. This implies dynamically inconsistent preferences. Second, we calculate an indicator of the degree of dynamic inconsistency that may help us to understand how households overcome their self-control problems. We use a micro dataset containing households’ reports on the compensation for receiving hypothetical rewards with delays. We find that individuals with more severely dynamically inconsistent preferences on average hold a statistically significantly lower share of their total wealth in checking accounts. A possible interpretation is that subjects use precommitment strategies to limit their temptation to consume immediately.

Also please please click here for the full paper.