The Concept of Time Preference is Completely at Odds With Reality

time preference

A conversation that I was having yesterday reminded me of a rather funny point in economic theory. When we consider the value of a financial asset we take into two components: that is, it’s price and it’s income stream. It’s price is a sort of stock variable while it’s income stream is a flow variable.

Now that’s all rather simple and elementary. But once we subject these two variables to some degree of uncertainty it becomes impossible to truly calculate the value of the asset moving into the future in any meaningful way.

Let’s take a concrete example: that of a government bond. Let’s say that the bond is worth $1,000 and is paying 5% interest a year and must be redeemed in ten years. If time was completely homogenous and the future was identical to the present I can calculate whether I should hold this bond vis-a-vis another financial asset rather easily. The same is true if I know that the future is going to change and in what direction it will change. But if the future is uncertain it becomes well-nigh impossible — at least in any fully rational manner.

I don’t know what will happen in the market between now and the time I want to sell the bond. It’s value might rise substantially vis-a-vis other assets or it might fall. All I can really do is take a punt on it and hope that I have an edge on the other guy; hope that I noticed some trend developing that he missed.

Okay, well that’s all very obvious, right? But now consider the way economic theory tends to think about rates of interest and so forth. James Tobin discusses this on a nice paper entitled Commentary on Irving Fisher, The Nature of Capital and Income reviewing Irving Fisher’s early book on the nature of income and capital. His first example is quite illuminating in this regard.

Would you rather live in economy J or economy U, when U is currently producing and consuming more but J is building more capital facilities and is growing faster? Samuelson’s conclusion is in the spirit of Fisher and of Hicks. Do not compare current incomes on any definition. Instead, measure welfare as the discounted value of the expected consumption stream of an individual or an economy. (p8)

Note already the dreaded word ‘expected’ rears its ugly head. Samuelson and the other marginalists merely say that we should ignore present income and instead make a judgement based on future income. Not only would I argue that this is not the way many people behave but I would also argue that it is not particularly rational because we cannot know the future.

Take an example. Imagine that I am living in London at the moment. I have a free ticket to move to Paris. Real wages are higher in London and all other conditions are identical in both cities, except that house prices are rising faster in London than in Paris. Even though I think that I’m fairly good at spotting housing bubbles I am not sure at all whether there is actually a housing bubble in London, whether it will burst or whether it bursting will have a substantial effect on either my wages or my employment situation. I have quite simply no way of accounting for this. (Although most of this is fiction the situation with property prices in London at the moment is precisely as I say it is).

In such a scenario — which is not totally unrealistic — I have no way of “discounting my future consumption streams” and the decision that I would likely make would be to stay in the high-income city with the rising house prices and take my chances. Indeed, any attempt to try to measure future events on which I cannot even place a realistic numerical probability would be, when all is said and done, a bit deranged. I might as well take up astrology or fortune-telling. To spin such activity as being ‘rational’ is really a bridge too far and actually an affront to what I would consider actual rational discourse.

This goes to a more immediate problem, especially in financial theory. Tobin writes of Fisher’s book:

[The] most important [idea in the book is] that the value of an asset is the capitalization of the stream of future services thrown off by the asset. (p9)

He then goes on to say that in Fisher’s later work this idea would become the basis for “the equilibrium condition for determining interest rates” — in contemporary parlance Fisher would go on to construct a time preference theory of the interest rate (give up consumption now for more consumption in the future) and the equilibrium condition would be the point at which society would maximise its utility at a given moment in time.

But look at the problems this throw up. First of all, how do I measure the “value” of the future services the asset will throw off? I do not know what I will want two days from now, let alone six months from now. An expensive t-shirt is far more ‘valuable’ to me if we have a hot summer than if we have a cold and wet one but there is no way on earth I would base my purchasing decisions on weather forecasts 3-6 months out! Thinking that I can somehow determine how “valuable” the services given out by an asset are in the future in any exact way is as ludicrous as trying to make the decision based on some quasi-numerical calculation as to which city I should work in. What if I get bored of the service the asset produces, for example?

More important than this, however, is the indeterminacy surrounding the price of the asset in question. How can I compare the services of, say, a car or a house to other goods when the relative prices will change in the future in line with (highly uncertain) asset price valuations? For example, if my house is worth £500,000 today I can in some very limited sense compare the “value of its services” with other goods in the economy. But what if the market collapses tomorrow and it falls in value to £250,000? How can I try to calculate the future opportunity cost of not holding another asset when all assets are subject to highly uncertain price dynamics? Simple answer: I can’t. I can only take a punt.

I have a pretty economical mind. I like to think I’m pretty good at making financial decisions and making money. And given that I have a fair grasp of macroeconomics and tend to be able to spot trends I probably have an edge over a substantial amount of the population. If I can’t make these decisions in a perfectly calculating manner then how on earth can economists try to build their theories of how the economy functions based on the idea that everybody in the economy is doing such calculations?

The fact is that anything such a theory produces — any ‘results’ — will only be so much dross. They will have no relevance to the real world whatsoever. Not in this universe, anyway. And that is why the Keynesian idea of uncertainty literally destroys anything resembling mainstream economics — whether micro or macro. Robinson once quipped that Keynes hadn’t taken the twenty minutes necessary to learn the marginalist theory of value. I don’t know if this was actually true but it it was, it would be quite understandable. He had already constructed his theories of probabilities and studied the money markets in depth. What nonsense the old marginalist theory must have looked like to him!


About pilkingtonphil

Philip Pilkington is a macroeconomist and investment professional. Writing about all things macro and investment. Views my own.You can follow him on Twitter at @philippilk.
This entry was posted in Economic Theory, Statistics and Probability. Bookmark the permalink.

29 Responses to The Concept of Time Preference is Completely at Odds With Reality

  1. Lord Keynes says:

    A great post.

    “The fact is that anything such a theory produces — any ‘results’ — will only be so much dross. They will have no relevance to the real world whatsoever.”

    Marginalist theory assumes people have some god-like omniscience in their ability to see the future: this is the only way it makes any sense. But the assumption is so stupid and wrong, it takes one’s breath away.

    • The irritating thing is that this is such a simple point to make. It’s something that any smart undergraduate can make. I would imagine that many do — to themselves at least before they drop the module. And yet if you raise it with a marginalist they will just shrug it off. They will not even answer you. They will look at you as if you’re being irrational and stupid and not try to justify what they are doing. It’s completely bizarre. I know of no other discipline in which an academic would get away with this.

  2. David Lentini says:

    Enjoyed the post very much. But I think things are even worse. We should factor in the affects of advertising and marketing, which typically try to convince the present potential buyer or investor of a certain future. Economics becomes even more deranged when we try to account for the noise created by those trying to convince us to buy or sell—-We really don’t make decisions in a vacuum; the “market place of ideas” is really about ” the marketing of ideas”.

  3. NeilW says:

    Discounted cash flow was all the range in the mid 1990s. It bit the dust after the dot-com bubble burst.

    I’m not at all sure you can capitalise anything in a meaningful way – as the variable value attributed to pension schemes in corporate accounts testifies.

    And certainly once you try and tie it down to real stuff, the whole process is no better than rolling a dice.

    Which is why, prior to financialisation, the focus was upon the income stream and the capital part was essentially a write off as a ‘sunk cost’.

  4. ivansml says:

    “About these matters there is no scientific basis on which to form any calculable probability whatever. We simply do not know. Nevertheless, the necessity for action and for decision compels us as practical men to do our best to overlook this awkward fact and to behave exactly as we should if we had behind us a good Benthamite calculation of a series of prospective advantages and disadvantages, each multiplied by its appropriate probability waiting to be summed.” – J. M. Keynes

    • Yes, Keynes the ironist trying to get through to his deranged colleagues. But let’s even take what he’s saying at face value.

      Then we are supposed to extrapolate from this that everyone else thinks and behaves in this manner, despite the fact that their IQs will vary across the spectrum and most will have never read Bentham or heard of ‘utility’? And then we’re supposed to extrapolate further still that all of these people acting in an ‘as if’ manner will converge on correct probabilistic assessments with regard to time preference and so forth?

      After swallowing that — it was not easy, even for you I’m sure — we then build a highly simplified model in which this all occurs at an instantaneous moment in time thus producing an optimal rate of interest… and then we infer from this model facts about the real world?

      Frankly that sounds like something a raving lunatic has scrawled on the wall of a padded cell while ranting that he has seen into the future and knows what will happen if people do not mend their ways and behave better. No, I think that I’ll opt out of the marginalist madhouse, thank you very much. I won’t be checking in today!

      The more I think about it the more I must conclude that this socially sanctioned delusion must be generated by some emotional need for power or control or something of the sort. It simply cannot be explained otherwise.

      • NeilW says:

        It is the job of the religious class to justify the actions of their benefactors.

        That’s how we ended up selling indulgences.

      • I can’t just put it down to the compensatory aspect, Neil, although grants and tenure and professional recognition obviously play a large part. (Just look at the, urm, intellectual development of a certain well-known young economics blogger who started off questioning modelling and ended up sympathetic to ‘innovations’ in DSGE modelling after some of the big players were nice to him at a party…).

        But it must run deeper than that. The manner of thinking must appeal to something in the lizard part of the brain. I think its control and power. I think its the same emotional appeal that horoscopes and astrology have.

      • ivansml says:

        It is one thing to criticize simplistic representative-agent, rational-expectations model (because that’s what you are actually doing). It is another thing entirely to deny that expectations matter at all, in asset markets and elsewhere. Sure, in reality the economy is characterized by uncertainty that’s unlikely to be literally described by objective probability measure. Yet in the end, people must make decisions somehow, and they *do* take into account expectations of future events.

        Moreover, they may have heterogeneous expectations formed by various heuristics and conventional judgment, subject to waves of optimism and pessimism, just like Keynes claimed. But that doesn’t mean that features explained by simpler models that ignore these elements (like Fisher’s explanation of interest rate by intertemporal substitution) can be just ignored. To argue so is simply non-sequitur.

      • Expectations are what drive investment and output in the economy. They are also — and this is my main focus right now — what drive prices across the economy; not just in asset markets, but in all markets. In a very real sense, the economy is merely a collection of expectations. So, you won’t find me denying anything like that. It is central to everything I consider important in economics.

        Here’s the rub though: can these expectations be quantified? Absolutely and categorically not. Can they be ordered — i.e. ordinally quantified rather than cardinally quantified? Again, no. Are they independent of each other? No, they reinforce one another and are, to use the old language, dialectical (think of the Keynesian beauty contest in this regard).

        If we accept all that there is nothing left of marginalist analysis. People cannot order their preferences either ordinally or cardinally. That eliminates utility theory and any theory of time preference. People cannot integrate accurate probability measures into their investment decisions. That eliminates most of contemporary financial markets theory and overturns the idea of a ‘natural rate of interest’. And so on and so on.

        Marginalist analysis is wiped away if we accept those basic propositions.

      • NeilW says:

        The ‘psychics’ still pack out the theatres – and that’s despite making drop offs on a regular basis.

        If you’re a supposed psychic and your high profile client dies in a horrible car crash shortly after consulting you, then you’d think that would be the end of your career.

        Apparently not.

        Bullshit people want to believe always finds a ready market.

      • ivansml says:

        “So, you won’t find me denying anything like that. It is central to everything I consider important in economics.”

        Not what one gets from your post, but ok.

        “Here’s the rub though: can these expectations be quantified? Absolutely and categorically not. Can they be ordered — i.e. ordinally quantified rather than cardinally quantified? Again, no.”

        Just because there’s no magical machine that reads people’s minds doesn’t mean that they don’t rank likelihoods and results of possible future outcomes when making decisions, qualitatively and quantitatively. Clearly, any model of individual decision-making then must describe expectations by ordinal or even cardinal variables, even if those variables are themselves unobservable and volatile. And one can always try to obtain proxies for expectations through surveys or market prices.

        “Frankly that sounds like something a raving lunatic has scrawled on the wall of a padded cell while ranting that he has seen into the future and knows what will happen if people do not mend their ways and behave better. “

        In addition to still mixing concepts up (again: rep. agent rat. exp. is much narrower model than subjective expected utility, as formalized e.g. by Savage), you apparently think that when we write down an expected utility function, we consider it a literal description of what people do inside their heads, summing together discounted expected utilities in some internal mental spreadsheet. Newsflash – we don’t. All formal models are just stylized analogies that hopefully capture some features of the world in a tractable way. Nothing more, but also nothing less.

        “The more I think about it the more I must conclude that this socially sanctioned delusion must be generated by some emotional need for power or control or something of the sort.”

        One could also speculate about motivations of those who prefer to write snarky “rebuttals” that are then circulated around their little heterodox corner of internet in a kind of echo chamber, instead of engaging actual (as opposed to caricatured) arguments and theories of the other side. But given your fondness for psychoanalyzing those who disagree with you, I’ll leave this line of inquiry to you.

      • Sorry ivansml, people just don’t think like that. End of story. When you try to model them like that all you capture is a caricature. And when you try to apply the results of your model to the real-world you produce only nonsense. If you engage in some introspection — terrifying prospect, I know — you will see this quite clearly. As I say in the piece: how can you rank your preferences through time hierarchically if you don’t know the future? How can you calculate the value you will derive over the next year from a pair of shorts when you don’t know the weather, for example?

        My motivations are obvious: I think that the marginalists are engaged in intellectual fraud. I try to point that out as best I can. To my mind the question then becomes: since I do not believe that these people are consciously lying how do they convince themselves of the truth of something that is so obviously wrong? I think that must have something to do with the “kicks” they get out of thinking that they have a theory of how people behave. It is the same emotional appeal as fortune-telling or astrology.

      • By the way, ivansml, I don’t believe that I am misrepresenting the other side. But perhaps I am. So, how about this: you provide me with a paper that you think is representative and I will write a post criticising it.

        Provide a link here and I will write a post in the next few days. If you’re willing to put your money where your mouth is, of course, because I would imagine I will find the same problems I am highlighting here in the paper you provide. But let’s see, shall we?

      • ivansml says:

        “So, how about this: you provide me with a paper that you think is representative and I will write a post criticising it.”

        Sounds like fun. But there are tens of thousands of papers (of which I know just a small fraction, of course), so you may want to narrow it down a bit – representative of what exactly?

      • You wrote:

        …instead of engaging actual (as opposed to caricatured) arguments and theories of the other side.

        Therefore I assume that you think that I don’t engage with your marginalist theories and have not done so in the above post. So, provide a marginalist theory and I will look at it. Maybe one tied to the supposed caricature I engaged with above (presumably Fisher and Tobin are the caricatures in this case…).

      • ivansml says:

        There’s no need to discuss Fisher or Tobin in particular, as they serve just as placeholders for “marginalist” theory in your post. But if you want to know what actual, living decision theorists think about probability and uncertainty, you can start with Gilboa, Postlewaite & Schmeidler (2008) [1], who provide nice, nontechnical summary of subjective expected utility, as well as of its limitations. And while you’re at it, check out “Economic models as analogies” by same authors [2], which discusses the role of models and theory in economics.



      • This has nothing to do with the above post. Indeed, glancing at the conclusions of the first linked to papers suggests that the authors would agree with my assessment of models that utilise time preference.

        Also, if you think that models are analogies then I assume that you are against econometric testing of models, right? Or do you just say that they are analogies when you are called out on BS and then in your day-to-day life use econometric testing to “prove” validity?

      • Also, if models are analogies and people are not Bayesian — that is the message of the two papers taken together — then aren’t models that pretend people are basically Bayesian bad analogies? I’m really struggling to see what you’re trying to say here…

      • ivansml says:

        Since your post can be basically summarized as “no one knows the future, so all those models are nonsense” (and note that specific example of Fisher’s theory of time preference is irrelevant here), I think the papers I linked are quite on point.

        Theoretical models are analogies. Econometric estimation, on the other hand, is mostly about the other branch of economics, i.e. applying those analogies (which motivate the formulation of statistical model, choice of variables, etc.) to study particular applications. On yet another hand, past empirical estimates may also serve as analogies for new applications. In any case, I don’t see a contradiction here.

        Yes, Gilboa questions the Bayesian paradigm, seeing it as special, not exclusive, case of rational behavior. But it can still be the case that allowing for more realistic models, with ambiguity aversion and other stuff, will not change our conclusions much. The only way to find out is to try and build such models, then apply them for studying particular research questions. Which is exactly what a lot of economists are doing.

      • So you admit that people cannot know the future. You provide me with a link that supports this case. And then you say that models that pretend that people know the future are fine so long as we add a few new variables.

        Well, I think that’s deranged. But let’s run with it. Please provide me with a model that you think is a “good analogy” that might be worth testing against the data and I will criticise it like I said.

        Ball is in your court.

  5. Lars Syll says:

    A great post, Phil. And judging from the comments it seems as though some guys — finally — have learned something about the status of “as if”- modelling …

  6. philippe101 says:

    intuitively ‘time preference’ seems to have something to do with interest rates.

    Say I have some money, I can decide to tie it up for a while in a savings account or I can choose to have immediate access to it in a current account. The former pays more interest… pretty simple really.

    Yes there is uncertainty about the future so people use rules of thumb and best guesses, but none the less I will demand more interest on a savings account because of my ‘time preference’, no?

  7. GermanAngst says:

    Phil, have you read about Tuckett: ?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s