How Does A Money-Printing Central Bank Lose Money?

The Swiss National Bank (SNB) has just announced an eye-popping net loss for the first quarter of 2015: 30 billion Swiss francs, or $32 billion[1]. A participant in its recent shareholders meeting shortly before the announcement told me “the directors looked very stressed.”

How does a money-printing central bank lose money? In this case, the SNB had 41 billion in losses on foreign exchange positions, primarily on its investments denominated in euros. These were bought to try to hold down the appreciation of the Swiss franc against the euro, which in the end didn’t work. The losses were partially offset by 10 billion in price gains on stocks and bonds in its portfolio. These were the dominating factors in the net loss of 30 billion.

Interestingly among smaller items, the income of the SNB includes 236 million in negative interest rates paid by depositors for holding deposits with it—a revenue source that American banks can only dream of.

The big loss is the result of marking assets to market—that is, using the most current market price, as opposed to “historical cost” accounting, which shows what you paid for an investment when you bought it, not what it is worth today. In its published financial statements, the SNB is required by Swiss law to mark its investments, which include its large securities and gold holdings, to market. The resulting financial statements, the SNB formally declares, “present a true and fair view of the financial position.” No American firm has to come forth with such statements, nor does the Federal Reserve.

After booking the 30 billion quarterly loss, the SNB as of March 31 posted a net worth of 56 billion, compared to its 581 billion in total assets. This gives it a 9.7 percent capital to-assets ratio, after the big hit.

In dollars, the SNB’s capital is now $60 billion. This is somewhat larger than the capital of the Federal Reserve, which is $58 billion. The capital supports assets of $622 billion in the case of the SNB, but $4.5 trillion at the Fed, so the Fed’s capital ratio is a mere 1.3 percent, only 13 percent of the SNB’s. Some other interesting contrasts between the two central banks:

The Constitution of the Swiss Confederation requires the SNB to hold part of its reserves in gold. The SNB holds 39 billion in gold, marked to market. The Fed owns zero gold. (People are sometimes confused by the fact that the Fed owns “gold certificates,” but these are only dollar claims on the U.S. Treasury—they certify that the government took the Fed’s gold, along with everybody else’s, in 1933 and gave it the certificates in exchange.)

Shares of the SNB trade on the Swiss stock exchange and, as noted, it has an annual meeting of shareholders, at which they ask questions of the management about the operations of the bank. In the United States, shares of the various Federal Reserves Banks—there are no shares of the Fed as a whole—are held by commercial banks and do not trade. There is no annual meeting of shareholders.

The contrast in accounting practices is even more interesting.

Would the Fed ever be as honest about a loss of value in its assets as the SNB is? Nope–not under Federal Reserve accounting rules, which the Fed sets for itself. The Fed books its vast investments, now $4.4 trillion of them, at cost. The Fed insists that it has a “unique nature” and marking investments to market would not be appropriate for its special status of being the central bank. It would certainly be unseemly to write them down! So if the Fed did have a big depreciation in market value, as the SNB did, you would not see it in the Fed’s balance sheet or profit and loss statement.

Is such a depreciation in asset values possible for the Fed? Oh, yes. In previous days, when its investments consisted primarily of short-term Treasury bills, it would not have been. But now the Fed is a big-time, unhedged speculator in $2.5 trillion long-term bonds and $1.7 trillion of long-term mortgage securities, whose market values can change a lot, and will significantly depreciate if interest rates go up.

The Fed does not report the duration of its investment portfolio. But a not unreasonable guess might be a duration of six years. That means a 1 percent increase in long-term interest rates would decrease the market value of the Fed’s portfolio by 6 percent, or about $268 billion. This would be about 4.6 times the Fed’s capital. A 2 percent increase in rates would approximately double this effect, to $537 billion and nine times the capital.

In such a case, one can imagine that Federal Reserve directors might look a little stressed, even without having to face an annual meeting.

The accounting of the Fed is even more “unique” than this, however. A mark-to-market loss on existing assets is by definition an unrealized loss, which might be reversed in time. How about a realized loss? Suppose the Fed actually sold some of its bloated portfolio at a loss—how would its reporting of that compare to that of the SNB? A lot less honest, is the answer.

Under changes to Federal Reserve accounting policy adopted in 2011, when it was ramping up its investment portfolio and realized it might experience losses, any realized losses the Fed takes do not reduce its book capital, but are hidden in an intangible asset account. (This, by the way, is what the savings and loans did with losses on mortgages in the 1980s thrift crisis.)

Suppose the Fed had a net loss of $58 billion resulting from the sale of underwater bonds and mortgages. You would think that would wipe out its capital. That would be the accounting result for anybody else. But in this case, the unique Fed would still report capital of $58 billion. If it had realized losses of $100 billion, it would still show capital of $58 billion—with losses of $200 billion, capital still $58 billion. So fine a thing it is to control your own accounting policy.

What must the serious Swiss think of all this?