Natural Rate of Interest

A manufacturer considering a loan needs to know what the bank will charge. An economist can ask another question: what interest rate would be consistent with equilibrium across the economy? Knut Wicksell sought a rate that would not itself push the general price level up or down. He linked it to an imagined world where people lend actual capital goods, such as production machinery, rather than money. This is a conceptual comparison, not an undiscovered loan offer waiting to be found. [1]

The modern Laubach–Williams (LW) and Holston–Laubach–Williams (HLW) models published by the New York Fed use data on output, inflation and interest rates. From their relationships, they infer an unobserved real rate for an economy making sustainable use of its workforce and productive capacity, with stable inflation. No instrument measures it directly. A model is a simplified account of how the economy works, so different assumptions can produce different results. [2]

Take a simple example of our own: a loan has a nominal annual rate of 4% and expected inflation is 2%. Its expected real rate is approximately 2%. The nominal rate tells us how much extra money must be repaid; the real rate accounts for what that money will buy. We have converted one rate, not discovered r*. If actual inflation differs from expectations, the real outcome calculated afterwards will differ too. [3]

“Short-term” can be confusing. It describes the lending period, not necessarily how far ahead the model looks. Stanley Fischer explains that the Laubach–Williams approach seeks a level after shorter-lived influences, including the business cycle, have worked themselves out. A model can therefore estimate the longer-run equilibrium level of a short-term rate. That is not a forecast of the rate a bank will offer next week. [4]

Imagine looking today at a chart of last year’s r*. Economists at the time may not have seen that same curve: some data did not yet exist and other figures were later revised. The New York Fed therefore distinguishes estimates based on vintage data from current model estimates. Judging an earlier decision requires knowing what was available then. A newer chart does not necessarily mean someone overlooked the original information. [2]

A 2020 Federal Reserve study shows that changing how inflation expectations enter a model can alter both estimated r* and its uncertainty. Expectations are not the same as price rises already measured. A point estimate expresses the result as a single number, without guaranteeing its accuracy. An uncertainty interval indicates the range of values the chosen method allows for. A negative point estimate does not mean economists directly observed that level either. [3]

Comparing an appropriate real rate with estimated r* helps assess whether monetary conditions tend to encourage or restrain spending. Both sides of the comparison must be meaningful, however, and the estimate remains uncertain. In 2018, Jerome Powell emphasized looking at multiple indicators. A gap between two numbers alone cannot establish whether a particular rate decision was right or wrong. [5]

You will encounter r* in discussions about rates and the economy that can also affect financial markets. It is not a forecast of Bitcoin returns. Neither a rise in its price nor interest offered by a service holding your funds measures r*. Such a comparison would also mix currencies, periods and risks. The concept helps you understand interest-rate debates; it does not give you a ready-made answer about what to buy. [2] [3]