When Goldilocks fed the bears

In a world where short-term interest rates are rapidly approaching zero, with the U.S. Federal Reserve mimicking Japan, and Britain and the European Central Bank expected to follow, the Bank of Canada’s successful approach of inflation targeting has reached a crossroads.

Ironically, the BoC, along with other central banks, may be the victim of its own success. What is now regarded as the “Great Moderation” of the past two decades — a Goldilocks economy, neither tepid nor on the boil — played a hand in the inflation of two massive asset bubbles. But the role was indirect. It’s not so much that low short-term interest rates encouraged excessive borrowing, but rather that a tame environment allowed investors and households to mis-measure the risk they were taking on: to borrow more in the expectation of ever-higher profits.

And the current approach of minimal overnight rates — while necessary to bail out the financial system — runs the risk of repeating that scenario and setting up another threatening asset bubble.

That’s the view of a recent C.D. Howe Institute commentary by David Laidler and Robin Banerjee.

The authors argue that inflation-rate targeting should not be abandoned; instead, more care must applied to targeting the upper bound of the inflation range (currently 3%). And, when the monetary policy agreement between the Bank of Canada and the finance minister comes up for renewal in 2011, Laidler and Banerjee recommend that the BoC’s role as lender of last resort be explicitly acknowledged.

Low inflation has been a boon for the economy. Yet, even at the time of the dot-com bubble — during the so-called Goldilocks economy of the late 1990s — some bankers spotted signs of fragility in asset markets. The Bank for International Settlements (BIS), Laidler and Banerjee recount, expressed concern that low inflation had caused investors to underestimate the risks they were actually assuming. And, as the bubble burst, the BIS foresaw more instability flowing from monetary easing. And so it came to pass.

Two policy implications follow from this. The first is that banks have to be concerned about asset bubbles. The second is that, if bubbles collapse, “monetary authorities ought to be wary about the amount of support they then provide … to the financial system in its wake, lest this encourage even more foolhardiness,” argue Laidler and Banerjee.

It would seem the experience of the U.S. bears this out, since the Federal Reserve did little to rein in the stock market bubble of the 1990s, and then, in lowering interest rates, fostered a housing boom.

How did this happen? Laidler and Banerjee assert that, with the relatively smooth economic expansion of the 1990s, it was easy to forget that there are business cycles, and moreover, that a business cycle is also a credit cycle.

Unlike the U.S. Fed, which must target both prices and employment, the BOC sets a band for inflation, aiming at 2%, with a band of 1% to 3%. During the Great Moderation, inflation behaved and real output grew steadily. Not so with asset markets.

This, say Laidler and Banerjee, is not at all surprising. Indeed, it’s very much like the pre-1929 bust in the U.S., and the 1980s experience in Japan, periods where asset bubbles formed despite restrained price increases. But in Japan, inflation did begin to rise in 1987. This is similar to what happened in the U.S., from the collapse of Long-Term Capital Management until the tech peak in 2000, and again after the stock market bottom in 2002 and until the asset-backed commercial paper crisis in August 2007.

Canada has had a better experience than the U.S., the authors argue, because it tightened monetary policy before the U.S. did, guided by its formal 2% inflation target. While the U.S. central bank has an informal target similar to Canada’s 2% goal, Canada targets consumer price inflation; the U.S. bases policy on the core personal consumption expenditure (PCE) deflator.

The PCE deflator takes into account not just changes in costs, but changes in their weight in a basket of consumer goods. As a result, it tends to give a lower inflation reading than CPI, which uses a basket of goods that are fixed in proportion. The PCE deflator acknowledges more quickly that consumers, faced with higher prices, will substitute cheaper goods and services. The problem, as Laidler and Banerjee define it, is that the U.S. Fed uses the “core” PCE, which strips out food and energy costs. The danger is that this measure overlooks a longer-term trend toward higher food and energy costs, costs that, in the short term, do exhibit volatility.

A similar mis-measurement has occurred in the U.K. While following an inflation-targeting regime akin to Canada’s, the price index does reflect food and energy, but not all the facets of housing costs. A better index would have been the U.K.’s retail price index, which is similar to Canada’s CPI.

But is using the proper price index enough to squelch asset bubbles? Laidler and Banerjee suggest not. Asset bubbles also occur in benign inflationary periods. In the dot-com period, for example, not only did stock prices run up, but so did output, with the result that companies were supplying credit to their customers. In the aftermath, there was a glut of product. Similarly with the housing bubble, not only did prices inflate, but that spurred a huge boost in construction; now there is a massive inventory of housing.

But this points to another factor. Innovation often spurs expansion in the hopes of greater profit. This occurs with technical developments that spur productivity, and with new lending arrangements such as securitization. Just when a sustainable boom passes the threshold into an untethered bubble, where expectations feed even higher expectations, is, unfortunately, visible only in hindsight.

Nevertheless, mismatches in the returns investors expect in particular sectors and borrowing costs can be at the source of inflation across the economy, as prices “play catch-up.” It isn’t always so, Laidler and Banerjee note, because credit creation and contraction in a given sector are subject to some random variation. But this suggests a closer look at the credit creation, not just conventional measures of money, but close money substitutes — novel short-term securities like asset-backed commercial paper — that do not show up in the conventional measures.

This poses challenges for monetary policy. Consumers and firms don’t use bank-set overnight interest rates in borrowing to buy goods and services; they use the rates prevailing in the economy, e.g., the prime rate (which banks have, for instance, not been reducing as quickly as they have in the past when the overnight rate has changed). In addition, it is not absolute borrowing that counts, but borrowing in view of an expected rate of return. As a result, steady inflation targets may encourage a contraction of bank or corporate interest rate spreads as the perception of risk falls; in that sense, steady money becomes easier money, setting the stage for asset bubbles and rising inflation. That doesn’t mean the current approach need be abandoned. But, say Laidler and Banerjee, “rigid rules for setting interest rates can become misleading over time, even where the task of policy seems to be the maintenance of already well-established inflation stability.”

There are limits to what a bank can do, however. The Canadian economy is not homogeneous. Yet the BoC’s overnight rate applies to all regions, and to all financial institutions, who in turn lend to all sectors. So the BoC cannot respond to overheated regions or sectors. That is more a role for regulators such as the Office of the Supervisor of Financial Institutions and the Canadian Deposit Insurance Corporation, the authors suggest.

Still, financial crises do happen, and the role of the central bank is to mitigate the damage “to innocent third parties.” But resolving them is another matter, since the line between liquidity and solvency is hard to draw. An illiquid institution is still solvent if it can sell assets; if central bank support is there, many sellers will remain solvent. Without that support, many sellers become insolvent and that, in turn, feeds into the real economy as firms and households are unable to roll over their loans. It all depends on the level of support.

“Thus, any financial crisis, even an incipient one, carries with it the threat of a downward spiral that involves the financial system and the rest of the economy. The quicker and more vigorous the central bank’s first response to trouble, not to mention the more confidently expected it is, the less likely is the spiral to get started.” But against that lies the risk of moral hazard, and the degree to which it occurs, Laidler and Banerjee argue, is established only in retrospect.

While a gold-standard banking system might have responded differently to moral hazard, today’s banking system has no such ready-made stabilizer. Still, in contrast to a gold-standard system, inflation targeting allows a generous provision of liquidity and lower interest to be complementary measures, not mutually exclusive during a period of financial crisis.

But there are risks of an over-expansionary response — as happened in the U.S. after 2001 — as well as of impairing public confidence in inflation targets that to date have been firmly anchored.

Laidler and Banerjee recommend that inflation targeting be continued, but that the de facto role of the BoC as the lender of last resort be explicitly acknowledged, to alleviate public uncertainly about the commitment to inflation targeting. And, to maintain public confidence, the relevant price index should remain headline inflation — food and energy should not be stripped out — because it is transparent and easy to understand.

It is harder, however, to target asset prices beyond housing costs, which do have a direct connection to the cost of living. Nevertheless, higher asset prices may be a sign of excessive credit creation, which can also show up in higher inflation. For that reason, Laidler and Banerjee argue that the upper band of the inflation target, 3%, should be treated as a hard limit.

(12/19/08)