2002 Japan Conference: A Summary of the Papers

The economics profession has been deeply divided on the issue of how to deal with deflation in Japan. Indeed, there has been great confusion and frustration at the inability of the profession to create a unified explanation and prescription. The explanation for such broad disagreement appears to be that the underlying models used to explain deflation differ widely among economists, and therefore the prescriptions differ as well. There are four such models that are the chief contenders in this debate. In addition, there are different models of policy formation that must be grafted onto the economic models in order to help us judge whether monetary policy has been correct.

The Monetarist Approach

The first model for deflation is the monetarist one. The Fisher equation (MV=PY) is its basis, and is restated in its modern form of M/P=kY. Next, a money multiplier process is added, so that base money (H), which the central bank controls, enters the equation. Thus, the usable equation becomes

mH/P = kY

The main contention of the monetarists is that H has not risen enough, and therefore that P is dropping.

The counter-arguments are both theoretical and empirical. The theoretical counter-argument is that with an unstable financial system, both the money multiplier (m) and the demand coefficient for real money balances (k) are unstable. Any rise in H is automatically countered by a drop in m or k. Hence, raising H will not solve the problem. The empirical argument follows from this. In fact, base money has risen by 80 percent since 1Q97, and the BoJ balance sheet by 120 percent. Broader aggregates also have risen, although only by about 20 percent. Hence, the multiplier is clearly unstable. In addition, the sharp acceleration of the BoJ balance sheet and of base money have been answered by equally sharp declines in the multiplier in recent years. The sharp shifts in the composition of broad money (from quasi-money into M1) also suggest that liquidity preference indeed has been unstable, in light of instability in the financial system.

Thus, if one asserts that even more aggressive increases in H would stop deflation, one must explain why, from this point onward at least, one expects the money multiplier to stabilize and/or the demand coefficient for real balances to stabilize. Without such an explanation, the assertion that "more H means less deflation" rings hollow.

The IS-LM Approach

Since the IS-LM model assumes prices to be fixed, it cannot be used directly as a tool for explaining deflation. However, if one's implicit model of prices relates changes in inflation to the degree of excess demand, then the key to stopping deflation clearly becomes demand management. The proponents of the IS-LM approach believe (in my view correctly) that Japan now faces a liquidity trap, and therefore that the LM curve is flat. Therefore, according to their view, the only alternative is IS policies. Thus comes the assertion that more government spending will end deflation.

The counter-arguments again are both theoretical and empirical. There are several theoretical objections. First, since the IS-LM framework has no explicit inflation model, the policy conclusion is only implicit. Second, the alternative effects of fiscal policy (for example, preserving excess supply) are ignored. Third, the effects of fiscal policy on the bond market also are ignored; this is particularly important in Japan today, when fiscal deficits are galactic, and jittery bond markets can explode (as occurred at the end of 1998) when adverse news is announced. The empirical objection is that fiscal expansion has been tried several times in the last decade, and has brought only temporary relief. Indeed, some commentators (myself included) believe that the temporary relief has dulled the edge of structural reform policies, which even the Keynesians agree are necessary.

The AS-AD Approach

The third approach uses the aggregate-supply (AS) / aggregate demand (AD) framework. The AD side is simply a restatement of the IS-LM model, whereby increases in demand correspond to lower price levels in equilibrium. (The reason is that increased demand requires higher real money balances; with a fixed level of nominal money supply, an increase in real money balances must come from lower prices.) Thus, the AD curve is downward sloping, with output on the horizontal axis and the price level on the vertical axis. The aggregate supply curve comes from the relationship between nominal wages (and other cost components) and the price level. At a given level of nominal wages (or interest costs, or oil prices), a higher level of prices would imply higher profits, and therefore more output. Thus, the AS curve is upward sloping relative to the axes just mentioned. The intersection of AD and AS curves generates an equilibrium. In order to raise the price level (that is, to stop deflation), one can use either the AD or the AS curve. This is where things get interesting.

Traditionally, the AD curve responds more to monetary and fiscal policy than the AS curve. (Indeed, the latter is assumed to be inert with respect to policy changes.) However, this is precisely the assumption that appears to fail in Japan today. Indeed, fiscal policy over the last decade was aimed largely at keeping inefficient producers (in construction and real estate) in business, thus artificially pushing the supply curve far to the right side of the diagram, while pushing the demand curve only slightly to the right. The small movement in the AD curve may be attributable to Ricardian equivalence, which pulled private demand inward when government demand moved outward, and low return on capital, which lowered the multiplier effects of government programs. Ironically, according to this view, both monetary and fiscal policies actually worsened deflation.

The counter-arguments to this view are chiefly empirical. While it is theoretically possible that the effects of expansionary monetary and fiscal policy could be perversely contractionary, the empirical evidence for the actual existence of this phenomenon is weak. There is anecdotal evidence, for example, the rapid recovery of the economy in Texas once the RTC began aggressive real estate sales. At an aggregate level, however, the evidence is thin. On the more practical side, any policy to stop inflation by shifting the supply curve inward would cause a massive recession and might not be politically sustainable.

The Trade Theory Approach

The issue of deflation can be approached from a completely different angle as well, that of trade theory. The approach focuses on the effect of changes in relative prices on the measured price index. For example, if tradable goods prices are falling, but non-tradable prices are not rising enough to offset this fully, then the aggregate price index is declining. This can occur even though consumer welfare is improving as a result of increased trade. In terms of the Heckscher-Ohlin model, a change in international prices can leave an economy producing inside the production possibility frontier when resource transfer is not smooth. The consumer welfare of the country actually can improve when the terms-of-trade change is positive enough to offset the waste of resources.

The challenge here is to link a trade model (which deals with multiple goods but only in relative prices) to a macro model (which deals with only a single good and absolute prices). When the essence of the economic problem is changes in relative prices, the use of one-good macro models will mislead. There is very little theory that addresses how macroeconomic policies should deal with distortions in relative prices. Indeed, the one-good assumption in the structure of macroeconomics makes it incapable of addressing the issue. The only possible solution is an implicit model that links unemployed resources (that is, the degree to which the economy is inside the production possibility frontier) to the deflation rate. Without an explicit model, however, the trade theory approach to deflation is incomplete.

There are several counter-arguments to the trade theory approach as well. First, the lack of an explicit linkage to deflation is a fault. Second, the absence of any nominal quantities in trade models makes the application of standard macro theory impossible. Third, the fact that one of the major distortions is in the land market means that one factor of production is mispriced, and not a good.

Policy Models

Most academics acknowledge that policy coordination is also difficult when separate bureaucratic interests are involved. The question for a monetary authority is whether to adopt a globally optimal policy that hinges on other actors doing their jobs optimally, when in fact the other actors are not doing so. This is a problem of game theory. Should the BoJ "do the right thing" (assuming that it knows the right thing), even if other actors refuse to do their parts?

The work of Robert Axelrod (in The Evolution of Cooperation, Basic Books, 1984) explores this issue. Axelrod posits an infinite prisoners dilemma game, and explores what strategies generate the best score against a broad range of other strategies, with a broad range of payoff matrices. The answer is that a "tit-for-tat" strategy is best. Players should cooperate on the first move, and the decision at time t should mimic the action of their counterpart at time t-1. This strategy does not always generate good outcomes. But it does generate the best average outcome. In subsequent work, Axelrod and other scholars isolated four criteria for successful strategies in game theory situations (see Duncan Watts, Small Worlds, Princeton University Press, 1999). The four criteria are that the strategy must be nice (it must not pick fights), retaliatory (it must retaliate against attacks), forgiving (it must stop fighting when the counterpart stops), and transparent (the counterpart must know how the player will react in any given situation).

In light of these results, the best policy appears to be retaliation in the face of exploitation by counterparts. For this reason, even though the BoJ has been quite accommodative, the BoJ's refusal to move to full inflation targeting has probably been a good decision.