The Liquidity Preference Theory of Interest comes from the Keynesian school of economics, and it forms the basis of Keynes' ideas about the demand for money in the economy. It also forms the basis of Sir John Hicks' LM Curve in his IS-LM Model of macroeconomic stabilization.
In a nutshell, the liquidity preference theory of interest is concerned with the desirability of different financial assets in terms of their liquidity and yield.
Naturally, any investor would like their assets to be as liquid as possible in order that they can be converted into cash as quickly as possible should the need arise. Cash itself is the most liquid asset of all, but holding cash comes at the cost of the lost earnings that could have been generated by investing in something that earns income i.e. a yield.
The ideal combination of liquidity and yield will differ according to each investor's preferences, but more of both is always desirable wherever it is possible. However, in reality, any efficient investment portfolio will necessarily need to sacrifice some of one to get more of the other.
The liquidity preference theory of interest rates came into being as an alternative to the flawed classical theory which sets interest rates as being determined by the supply of savings and demand for investment loans i.e. higher interest rates induce more saving but deter investment. Where the two curves intersect the market is cleared at a given interest rate just as in any other supply and demand model.
The classical model may still have relevance in the long run, when the economy is operating at full capacity and fluctuations in money markets have no permanent effects, but in the short run the money market certainly matters and a more complete model of liquidity is needed.
Keep in mind that the liquidity preference model comes from Keynes' general theory, meaning that the emphasis is on the aggregate demand side of the economy as a determinant of economic output, at least in the short term. Effective aggregate supply is assumed to be on the horizontal section of the curve with plenty of spare capacity for the economy to grow without causing inflation.
One of the problems with the classical model is that it fails to take into account the effects of the fractional reserve banking system i.e. credit/lending can be created out of thin air by the banking system such that the supply of money can be massively extended without any associated increase in interest rates. This is, of course, one of the primary reasons why the western world has racked up such enormous debts over the last few decades even though interest rates have actually fallen to near zero rates.
Contrary to the classical theory, the liquidity preference model correctly emphasizes that it is demand for money from borrowers, not supply of money from savers, that drives the credit market and determines interest rates. The supply of money is determined by the central bank and the fractional reserve banking system.
Keynes defines three components of liquidity preference i.e. the various demands for money:
The transaction demand for money is as simple as the name suggests, it refers to that proportion of income that people wish to keep on hand in order to make their daily purchases. An important implication here is that the liquidity preference here is for 'real money' not nominal money i.e. if prices change then the amount of money demanded in order to make our desired purchases will also change.
The transaction demand for money is usually expressed as a function of income, meaning that as people earn higher incomes then they will tend to increase the amount/value of their intended purchases, and so transaction demand will increase. Similarly, if income falls then the reverse situation arises. Note that this is consistent with Keynes' fundamental law of consumption - see my article about the consumption function for details.
Relating this to the LM curve it means that real balances (M/P) are the relevant variable. If this was not the case, and nominal money mattered instead of real money, then there would be an incidence of 'money illusion' whereby people mistakenly react to income changes without adjusting for the effects of price level changes.
It is difficult for consumers to know in advance precisely what purchases they will make from one period to the next, and because of that there will need to be some additional money kept on-hand in case of some unforeseen expenditure arising e.g. repair costs if something breaks.
The amount of money needed to protect against such unforeseen costs is not large but it is necessary. The preference for liquidity here is a sort of insurance against the inconvenience of being unable to pay for something that is needed. As with the other components of money demand, people will be reluctant to hold too much money on-hand because of the lost interest that it could be earning (although there is also the danger of theft, or losing the money in some other way).
For wealthy individuals there is a need to decide on the optimum way in which to store value i.e. how to invest in order to earn the best interest rates with the least risk. Optimal risk/return ratios will differ for different investors with some speculators happy to bear more risk for higher potential rewards, and other investors just looking for a safe but sufficient return on their investment portfolios.
Since the highest yields can be expected to incur the greatest loss of liquidity, the theory suggests that investors must balance their portfolios such that they can earn yields while still being able to readjust their portfolios as the need arises. Pouring all money into assets with a long tie-in period before it is possible to get your money out is another sort of risk, and it can actually lead to losses e.g. a long-term asset may look attractive today, but if inflation unexpectedly rises after you have sunk a large portion of your money into it, then the real rate of interest earned may end up negative.
Because of all the potential risks involved in investment, most portfolios hold at least some portion of their funds in very liquid assets, currency being the most liquid. Currency will of course be eroded by inflation since it pays no interest at all, but it is not tied-in to anything and can quickly be moved into some other form of investment if circumstances require it.
The speed with which investors adjust there portfolio preferences is subject to some delay because it requires a change to their future expectations regarding risk and return. Interest rates and inflationary expectations do not instantaneously adjust to changing macroeconomic factors, and some changes may be viewed as transitory. Of course, once those changes are known not to be transitory then larger demands for liquidity adjustments will occur.
Special mention has to go to the bond market because this is the market that really determines money supply growth or contraction (and by extension the interest rate). Government bonds (and commercial bonds) are bought and sold due to the speculative demand for money. Government bonds and short-term treasury bills dominate the market, and it is movements in the demand for these assets that dictate monetary outcomes.
Suppose that the government wishes to borrow more money; it will instruct the central bank to sell a new issue of government bonds. These might be valued at, for example, $2m each and pay an annual return of $80k each. The inferred interest rate here is 4%.
If the central bank raises $10bn from this sale of bonds then the money supply in the economy will be reduced because payment came from the private sector i.e. investors took money out of their portfolios and gave it to the central bank in return for the bonds. The overall reduction in the money supply circulating in the economy will be some multiple of $10bn because the banking sector now has less money to lend out. As explained on my page about the money multiplier, banks can lend the same base money multiple times, and so when that base money is reduced it has a multiplied reduction in overall money.
Now, if inflationary expectations rise, investors will be less willing to hold on to those bonds because their real expected yield will be lower.
If they choose to sell the bonds then they must lower their price from the $2m which they originally paid - this implies that the interest rate these bonds pay will rise, because the $80k annual return must continue to be paid. This in turn will have knock-on effects for the interest rate on other bonds, which must also rise unless the central bank is determined to maintain the original rate (which it can do if it buys up, at the full $2m original price, any bonds that investors want to sell).
E.g. if the bonds fell in price from $2m to $1.6m, the implied interest rate from the annual return of $80k would be 5% rather than 4%. In order to maintain the 4% interest rate, the central bank will need to agree to buy back the bonds at the original sale price, but in doing so it will boost the money supply and undo the original reduction of the money supply.
If, on the other hand, the central bank wants to get ahead of inflation by pushing interest rates even higher to 6%, then it may choose to offer another issue of bonds with a sale price of $2m each with an annual return of $120k each. Market forces would then push the earlier bonds down in value to a rate consistent with a 6% interest rate i.e. 80k/0.06 = $1,333,333 each. This would clearly represent a very significant loss for anyone who had paid the original $2m, and explains why bond investors are so nervous about inflationary expectations and are highly likely to have a higher liquidity preference when inflation pressures are rising.
The Liquidity Preference Model is arguably the most controversial concept to come from Keynes' general theory, but it is certainly a huge improvement on the classical model of interest, saving and investment.
The main alternative to it is called the Loanable Funds Theory, and I do think that while there is some overlap between these two theories, it is a significant improvement on Keynes work.