Oh, the trials and tribulations of being a modern central bank! There are the unrealistic expectations of how much you know about the economic present and future (encouraged by your own PR, of course). There are the equally unrealistic expectations of how much you can control the economic future (encouraged by textbooks). And then there is the infinitely frustrating nature of the financial markets which you would like to manipulate. For financial markets are densely recursive, complex systems of self-reflective interactions and expectations, including actions which reflect expectations of what you will do all the time. This makes them hard to forecast, for the central bank just as for everybody else.

The last 15 years of finance has brought us three main surprises-three huge bubbles and their subsequent collapses: the tech stock bubble; the U.S. housing bubble (plus notable housing bubbles in Great Britain, Spain and Ireland); and the now-collapsing European government debt bubble.

Such financial adventures do not mean the people involved were just stupid. On the contrary, some of the very smartest private and government actors contributed to the systemic interactions which created the bubbles. These included, among others, the central bankers.

Not only was the U.S. housing boom stoked by Federal Reserve actions, but transcripts released by the Fed make it clear that they entirely missed the magnitude of the problem of the housing bubble and failed to anticipate the coming crisis. This is in spite of diligent and constant economic forecasting by scores of intelligent and well-intentioned Ph.D. Fed economists, armed with big computers and databases, and earnest discussions of the risks and outlook by senior officials.

Here is John Maynard Keynes' clever analogy for the problem of predicting financial markets: Imagine a competition "in which the competitors have to pick the six prettiest faces from a hundred photographs, the prize being awarded to the competitor whose choice most nearly corresponds to the average preferences of the competitors as a whole; so that each competitor has to pick, not those faces which he himself finds prettiest, but those which he thinks most likely to catch the fancy of the competitors, all of whom are looking at the problem from the same point of view" — all, in other words, predicting each other's predictions of everybody else's behavior.

This recursiveness of financial markets gives rise to Goodhart's Law, which reflects that when officials try to use statistical relationships discovered by economists in order to control financial behavior, their actions in doing so change the behavior and the relationships. As stated by banking expert Charles Goodhart in 1975: "Any observed statistical regularity will tend to collapse when pressure is placed upon it for control purposes."

How different this real world is from the unfulfilled dream of making economics, central banking and finance into Newtonian, predictive, mathematical science. Using Newton's laws, astronomers can calculate with accuracy the paths of the planets for hundreds of years into the future. All experts will agree with each other in these predictions. But how far ahead can financial events be accurately predicted? Not even three months, and economists will always disagree with each other about them. The motions of the heavenly bodies and the booms and busts of financial markets are two different orders of reality.

As one essential element of the problem, take interest rates. Over time, interest rates display surprising behavior — indeed behavior that previous market participants considered simply impossible. Interest rates on 10-year U.S. Treasury notes in March 2012 were at about the same level as they were in March 1950 — 2 ¼%. Since last March they have gone even lower, although most professionals a year ago thought they had to go up. As one thoughtful investment manager recently wrote, "This remarkable bond market continues to be a surprise ... there is no question we are in extraordinary and bizarre times."

Six decades ago and now, the Fed was and is manipulating bond prices to keep long-term interest rates artificially low. In 1950, this manipulation had been going on since during the Second World War. How long can it go on now?

How low can interest rates go? How high can they go? In both directions, more than people think. An old colleague of mine worked on a major bank's strategic planning effort in the 1960s to imagine the highest which interest rates could possibly go. The professional bankers' answer: 6%. In fact, 10-year Treasury rates peaked at over 15% in 1981 — a level unbelievable in 1950 and 1960, and once again unbelievable now. The interest rates of the early 1980s were the death-knell of the savings and loan industry and the post-war American mortgage finance system. The Fed's current zero interest rates, if they keep on, will be the death-knell of savers.

Regulators in the wake of the 21st century crisis are busy designating large financial firms as "SIFIs" — "Systemically Important Financial Institutions" — ones which can create systemic risk. The history of interest rates shows us the vast power of the central bank to create systemic risk.

Neither the Fed nor any other central bank is in a godlike position above the system of financial interactions, looking down, and able to understand and decree accurately from on high. No one is outside the complex, recursive system. Rather, they are all enmeshed within it, like Keynes' photo pickers. This makes the Fed's mistakes easier to understand. As then-Senator Jim Bunning is said to have asked Fed Chairman Ben Bernanke, "How can you regulate systemic risk when you are the systemic risk?" — an excellent and unanswered question.

The Fed is the biggest SIFI of them all.

Alex Pollock is a resident fellow at the American Enterprise Institute in Washington, D.C. He was president and CEO of the Federal Home Loan Bank of Chicago from 1991 to 2004.