Gauti Eggertsson

During the past decade, the fall in worldwide inflation has raised interest in the risks associated with negative inflation, or deflation. What gave the issue an element of urgency was the long-tasting recession in Japan, which was accompanied by mild but persistent deflation and short-term nominal interest rates close to zero. In such an environment, cutting short-term interest rates, the usual remedy for weak demand, was just not an option. This survey reviews IMF research on how to counter weak demand at low interest rates.

The first response of many economists to Japanese deflation was to suggest that the Bank of Japan simply print more money. It was believed that such a policy, sooner or later, would reverse the deflationary trend and stimulate demand. Over the years, however, while the Bank of Japan was printing money at an unprecedented rate, short-term interest rates were getting closer to zero, and the effect of money supply increases was modest at best. Since 1997, the Bank of Japan has more than doubled the economy’s monetary base.

Considerable research at the IMF has focused on how such a doubling of the monetary base could have had no effect on prices. The answer dates back to Keynes: Japan has found itself in a liquidity trap.

Eggertsson and Woodford (2003, 2004) show that a liquidity trap arises when temporary shocks make the zero interest rate bound binding. When this is the case, government bonds and money become perfect substitutes, and “quantitative easing” (the injection of liquidity through open market operations in government bonds), if it leaves expectations unchanged, has no real effects and only makes individuals substitute money for bonds in their asset portfolios. Monetary policy can still be effective, but only if it is able to change expectations about future interest rates and inflation.

Eggertsson (2003) shows that if the central bank is unable to commit to future policies, and minimizes a standard loss function, quantitative easing has no effect on expectations. Indeed, the public expects that any increase in the monetary base will be contracted as soon as the zero bound is no longer binding. This credibility problem, and the associated inability of a central bank to increase inflation expectations, produces a “deflation bias.” The paper suggests several policies to solve this credibility problem, such as printing money and buying foreign exchange, and engaging in deficit spending. These policies credibly increase inflation expectations because they give the government an incentive to inflate in the future.

Jeanne and Svensson (2004) explore the same credibility problem but assume that the central bank is an independent identity. They find that printing money and buying foreign exchange can also solve the credibility problem as long as the central bank seeks to avoid capital losses. In this case, a successful intervention in the foreign exchange market, along with the appropriate manipulation of the central bank’s balance sheet, leads to a depreciation and a corresponding increase in inflation expectations.

Ramaswamy and Samiei (2003), from an empirical standpoint, look at the effectiveness of foreign interventions in Japan between 1995 and 1999, when the interest rate was close to zero. During this period, on several occasions, the Japanese government tried to intervene in the foreign exchange market in response to deflationary risks. The authors find evidence that exchange rate interventions were successful over this period to depreciate the yen (and to strengthen it when the authorities wished so).

The classic solution to the credibility problems discussed above is a rule-based policy, à la Kydland and Prescott (1977). In this spirit, Hunt and Laxton (2003) support the introduction of a price-level-targeting rule to stabilize the economy when the zero bound is binding. They find that such a rule has useful properties in a liquidity trap. Indeed, by allowing a central bank to credibly commit to a price-level target, any deflation is going to be associated with expectations of future inflation. Since higher inflation expectations are the way to reduce the real interest rate, the rule will be effective in boosting demand when the zero bound is binding.

Hunt and Laxton (2003) also explore the effectiveness of expansionary fiscal policy in periods in which the zero bound is binding and find that such a policy can largely eliminate output losses. Eggertsson and Woodford (forthcoming) come to a similar conclusion but consider the use of distortionary taxes as a policy instrument instead. On the empirical front, Kalra (2003) studies the effectiveness of fiscal policy in Japan from 1960 to 2000. Using a structural vector auto regression framework, he finds that short-term fiscal multipliers have remained relatively stable over the past years although they have declined over longer horizons.

The problem of excessive deflation has not been isolated to Japan. During the past several years, Hong Kong SAR has experienced deflation as well. N’Diaye (2003) and Schellekens (2003) study the experience of Hong Kong SAR with deflation. Both authors find that the main culprits have been cyclical shocks amplified by balance sheet and wealth effects. N’Diaye (2003) uses a structural vector error correction modeling approach, while Schellekens (2003) looks at the price dynamics between Hong Kong SAR and Shenzhen, a neighboring city in mainland China. Kumar and others (2003) study the danger of deflation throughout the world, with a comprehensive review of earlier work and several new findings. They find that the danger of deflation in the United States is fairly small, but it is higher in countries such as Japan, Hong Kong SAR, Taiwan, and Germany. Finally, Baig (2003) looks at deflationary episodes in Sweden and the United States in the 1930s and, from such a historical perspective, argues for more aggressive and sustained quantitative easing than what has so far been undertaken by the Bank of Japan.