Tuesday, 18 October 2016

Wealth Concentration and Loanable Funds

Jo Michell has an interesting post on loanable funds.  This was prompted by the question of whether wealth concentration has led to rising asset prices and falling yields.  Whilst sharing Jo's views in many respects, my own analysis here is slightly different, so I thought it worth setting out.

First of all, I would agree that increasing wealth concentration is likely to increase the propensity to save out of income.  However, this in itself is not enough to change interest rates.

The diagram below is the standard model of loanable funds (borrowed from Jo, and in turn from Nick Rowe).  The idea here is that if there is an increase in the desired quantity of saving at any given interest rate, this is represented by a rightward shift in the Sd curve.  This leads to a fall in the equilibrium interest rate (even if the Id curve is vertical).


However, an increased propensity to save is not the same thing as a desire to save an increased quantity.  In fact, an increased desire to save leads to lower income.  The actual quantity people end up wanting to save is unchanged; it simply represents a higher proportion of income.  This is the paradox of thrift.  So there is no movement in the Sd curve and no change in the (partial) equilibrium interest rate.

There may be temporary movements, if people are mistaken in what they expect will happen.  If one person tries to save more without realising others are also doing so, then they may overestimate their own income and start to bid up asset prices.  But, assuming they eventually cotton on to what is happening, this will eventually reverse.

Having said all this, I would still say that increased wealth concentration has contributed to falling yields.  The reasons for saying so are as follows.

The argument above is based on what happens in a monetary economy.  In looking at how things pan out in a monetary economy, we can't get very far without asking how monetary policy is framed.  For example, if monetary policy takes the form of simply fixing or targeting a particular interest rate, then almost by definition changes in savings propensity are not going to change that rate.

However, for our purposes here we need to recognise that current monetary policy takes the form of inflation targeting.  This means that the central bank responds to perceived deflationary pressures by reducing the policy rate.  So an increased propensity to save does indeed lead to lower interest rates, but it does so because it depresses demand and because the central bank reacts to that.

One interpretation here is to suppose that implementation of monetary policy brings the economy back to its original level of demand, at whatever interest rate that requires.  Thus we can go back to our loanable funds diagram and say that the Sd curve has indeed shifted to the right, once we have taken into account the full operation of monetary policy.

This is one way of looking at it and I'm all in favour of looking at things in different ways, for the additional insight it provides.  However, I find it a little problematic.  We can easily conceive of a situation where there is an increased propensity to save but where reductions in interest rates are ineffective in restoring demand.  In this scenario, there is no solution compatible with the unchanged output interpretation and it fails as an explanation of interest rates.

Even if we think that interest rate manipulation can bring output back to its original level, there is no general reason to suppose that the terminal interest rate of this process is independent of the path taken to get there.  (In most models it is independent, but that doesn't mean it is in reality.)  There would then be no meaningful ceteris paribus solution to the loanable funds model. 

There is another important way in which wealth concentration has contributed to falling yields, one that was particularly important in the run up to the crisis.  This is not to do with increased propensity to save, but rather with portfolio preference.  (We might think of portfolio preference as a more general form of liquidity preference, when we are considering a range of assets rather than simply bonds and money.)

Wealth concentration makes investors more concerned about large exposures to single name risk.  Pre-crisis, managers of large cash pools already holding sizeable unsecured bank deposits increasingly sought alternative low risk short term investments.  This created a strong demand for traded high quality assets which could be used as collateral for short term secured instruments.  This in turn led to increased demand for the assets that could used to create such collateral, such as securitisable mortgages.  The effect of this demand pressure was to drive down the yields on such assets relative to policy rates and rates on unsecured bank deposits.

In short, wealth concentration is certainly an important part of the picture of what has happened to financial asset yields.  But whilst the loanable funds model might provide some kind of insight it is, in my view, an insufficient framework for understanding the mechanisms at play.


  1. Hi Nick,

    Agree with it mostly. I also raised a similar point on Twitter with Jo Michell.

    I am interested in this:

    "However, an increased propensity to save is not the same thing as a desire to save an increased quantity. In fact, an increased desire to save leads to lower income. The actual quantity people end up wanting to save is unchanged; it simply represents a higher proportion of income."

    That depends on what kind of model you have - whose saving etc.

    So with government, a rise in propensity to save will although reduce income but might increase private sector saving. Total saving will fall because total investment will fall because of fall in demand and because of the saving-investment identity.

    1. Yes I'd agree with that. The actual quantity of saving is always determined by the quantity of investment, but it is possible that the quantity of investment will respond to savings decisions for reasons other than interest rate changes, such as in response to output changes. In that event, the quantity of savings, and indeed the quantity people end up wanting to save, will change in line with investment.

  2. The savings equal investment curves have never made sense to me as it applies to macroeconomics. The reason is that it flies in the face of how new money is generated.

    New money in the fiat realm is generated when loans are made by banks. New money can only be saved by definition. We can call new-money "investment" or we can call it "savings", but in reality it is "NEW-MONEY".

    Now on a micro economic scale, higher interest rates have an effect on decision making, influencing the propensity to create new money (by taking bank loans). Now we can look at the savings equal investment curve as representing two points of view, the borrower vs the lender. Now the curve makes sense but there is no shifting of the lines to the left or right UNLESS the curve is expanded to cover two time points--between which--NEW-MONEY has been injected into the macro-system. Such a NEW-MONEY event would be equivalent to changing the monetary scale on the bottom (savings-investment) line.

    1. We tend more commonly to talk about savings and investment, but if it means anything loanable funds is really about borrowing and lending, which is not the same. Money creation, when banks make loans, is a borrowing / lending matter. It needs one party who wants to assume a debt liability and another who is prepared to accumulate the new money as an asset.