Greg Ip from The Economist on NGDP Targeting

Columnist Greg Ip from the Economist criticizes of nominal GDP targeting. He claims it won’t provide a Volker moment. First, he argues that the change of regime from interest rate targeting to quantity of money targeting did not cause an immediate change in inflation expectations. He tells an anecdote, shared by Volcker, of a businessman who agreed to a 13 percent wage agreement right before the disinflationary policy. His point was that Volcker’s policy actually worked by crushing demand. The economy moved up the short run Phillips curve, with unemployment rising and inflation falling.

Market monetarists have not said that the Fed must simply shift to a new regime of targeting the growth path of nominal GDP, and then nominal GDP will shift to the new path. Many of us have argued that this is possible, but the other part of the program is lowering the interest rate paid on reserves to something slightly below zero combined with heroic open market operations—something like 25% of the target for nominal GDP. That would be $4 trillion. However, that is not an upper limit. Open market operations should continue until nominal GDP hits the target. The Goldman-Sachs proposal also includes a substantial increase in quantitative easing. Krugman, DeLong, and Romer presumably favor the same. DeLong recently proposed charges on reserve balances at the Fed, heroic quantitative easing, and nominal GDP level targeting--the policy trifecta that has been advocated by Market Monetarists for several years now.

GI argues:

If the Fed promised to hold rates low until NGDP returned to its pre-recession trend, that would no doubt help hold interest rates down. But there are many, potentially superior, ways to achieve the same thing, such as a promise to keep short-term interest rates at zero for a specified period of time, to target bond yields, or to keep rates low until a particular inflation or unemployment rate is achieved.

Market Monetarists don’t favor the “keep interest rates low until NGDP returns t its pre-recession trend.” Much of the work on the role of expectations in raising nominal GDP back to trend is not about how expectations would drive nominal GDP to target without the Fed doing anything. It is more about how nominal and real interest rates—especially long term ones, but even short term ones—should rise as expectations shift. Promising low interest rates or targeting bond prices (long term to maturity ones?) is wrongheaded.

On the other hand, some complicated inflation and unemployment rule is more like targeting the level of nominal GDP. Nominal GDP targeting is really about targeting the growth path of the flow of money expenditure on output. It is a nominal value and so something the monetary authority can sensibly target. How changes in nominal expenditure impact prices and output is left to market forces.

I presume that GI has in mind something like the Evan’s proposal. Keep interest rates low until either the unemployment rate falls to 7 percent or else the inflation rate rises to 3 percent. In other words, if the inflation rate rises to 3 percent, efforts to reduce unemployment will cease and interest rates will be increased. Or, if the unemployment rate falls to 7%, then interest rates will increase. If instead, the policy is to keep interest rates low until unemployment reaches a certain level, the results could be catastrophic—hyperinflation.

So, why 7 percent and 3 percent?

More importantly, AND THEN WHAT?

With a regime of nominal GDP targeting, once nominal GDP reaches the targeted growth path, it is kept there. Does GI really favor a monetary regime based upon estimates of the slope of the short run Phillips curve? What role does stipulated values for the “social utility function” play? (Or is it assumptions about the utility functions stipulated for representative agents in some model?)

GI continues:

Second, a nominal target should encourage firms and workers to behave in a way that makes the target self-fulfilling. This channel is well established for inflation targeting: if workers and firms believe the Fed will keep inflation at 2%, they will tend to set prices and wages accordingly. Exactly how an NGDP target would be self-fulfilling is unclear to me: I haven’t seen good empirical or theoretical evidence linking NGDP targeting to the behavior of private actors.

If the Fed is targeting a 5% growth rate and potential output is expected to grow 3%, then the expected inflation rate will be 2%. Workers and firms will set prices accordingly. There is no difference.

GI, however, never explained why workers and firms “set prices accordingly.” It is based upon a notion that the flow of spending on output will clear markets at that price trajectory. In other words, it is based upon the assumption by firms and households that nominal GDP will match that inflation rate plus growth of potential output, along with a perhaps gradual closing of any output gap between actual real output and potential output. Otherwise, the target inflation rate would result in shortages or surpluses making such price setting suboptimal.

Or so the theory goes.

Of course, nominal GDP does make a difference if potential output is expected to grow more quickly or slowly than trend. Or, if nominal GDP is away from the target and needs to readjust. Admittedly, there is less theoretical work about regimes where firms and workers have expectations about future price levels rather than rates of change. As for empirics, the gold standard was a system that created expectations about price levels rather than inflation rates, but it was a bit of a moving target. Most Market Monetarists look at the Great Moderation. Take one look at the growth path of nominal GDP. And then look at the inflation rate. This was an episode of the households and firms setting prices consistent with a 2% inflation rate? Or was it an episode where firms set prices and levels of output so that nominal GDP remained on a stable, 5.4% growth path?

GI argues that the Fed’s problem in 2008 was that it failed to forecast how much nominal GDP would slow. He says that the Fed forecasted that inflation would be 1.5 percent and real GDP would barely grow. That is nominal GDP growth of 1.5 percent, well below the trend growth path of the Great Moderation. But his point was that if the Fed had known that nominal GDP would actually be falling at an 8 percent annual rate, it would have done more. And while perhaps the Fed should have done more even with the less alarming forecast, it must worry about its lack of technical expertise.

The Market Monetarist view is that the reason nominal GDP collapsed is because firms and households believed that the Fed would fail to rapidly reverse the drop in nominal GDP. In other words, there was little confidence in a Federal Reserve that would talk about keeping short term interest rates low for an extended period because it expected inflation to remain low.

That leads to the next argument. GI can’t believe that expectations that the Fed will soon reverse deviations of nominal GDP from target will keep nominal GDP from falling below target. Incredibly, he then makes the statement:

My colleague makes an even more extreme argument: that the recession could have been avoided altogether had the Fed pursued an NGDP target. He argues that recessions could only occur because of real shocks, by which I assume he means supply-side shocks. Taken to its logical conclusion, this is like saying there is no demand shock, whatever the source—terrorist attack, asteroid strike, stock market collapse—that the Fed could not fully offset. Economic actors would tell themselves, “Ignore the asteroid strike. The Fed will ensure everything will be alright.” This is simply not compatible with theory or evidence.

An asteroid strike is a supply shock. A terrorist attack that was highly successful would be a supply shock. I suppose we can imagine a terrorist attack that doesn’t destroy anything useful, but causes people to save, or more specifically, hoard money and so it is a “demand shock.” Of course, a stock market crash that led to an increase in the demand to hold money would be a “demand shock.” We actually have had one of those that didn’t impact nominal GDP. And so, yes, I do think that demand shocks will be mild if nominal GDP is targeted. If people save more, and any increase in the demand to hold money is accommodated, then the market interest rate falls with the natural interest rate, and people who are less fearful or who didn’t buy into a speculator bubble find that they can make purchases at bargain rates.

IG then points to the instability issue:

An NGDP target has some advantages over an inflation target, especially in responding to supply side shocks. But it could dangerously complicate policy making in more normal times. As inflation rises, the Fed tightens to keep nominal GDP on track; output then falls, but then so does inflation; the Fed must quickly loosen again. In a model developed by Larry Ball in 1996, NGDP targeting produces systematic over- and under-shooting of both inflation and output. It is “not just inefficient, but disastrous. It causes both output and inflation to wander arbitrarily far from their long-run levels.”

I don’t believe this is realistic. That is, Ball did develop a model that has this consequence. His model of the macroeconomy was pretty standard. However, I don’t believe that it is realistic to expect inflation and real output to wander arbitrarily far from their long run levels when nominal GDP grows on a target growth path. Something is wrong with these models. At best, they don’t apply to a nominal GDP growth path regime. At worst, they don’t apply to any regime.

With inflation targeting, the price level just evolves based upon actual inflation. Actual inflation is influenced by expected inflation and actual nominal expenditure. There is no long run equilibrium price level.

With a target for the growth path of nominal GDP, the long run equilibrium price level is equal to the target value of nominal GDP divided by potential output. The expected price level is the expected value of nominal GDP divided by the expected level of potential output. If prices are set too high, it is because potential output is greater than expected. The result will be surpluses—real expenditure less than potential output. This motivates lower prices, or a slower increase in prices. If prices are set too low, it is because potential output is lower than expected. The result is shortages which motivate higher prices, or more rapid increases. Prices and output converge to their long run equilibrium values.

However, it is important to note that most of the changes in prices should occur in the particular sectors of the economy where potential output is higher or lower than expected. As Selgin has emphasized, this is not much of a problem. When there is a good corn harvest, the corn farmers must lower prices to sell their output. Potential output is higher than expected, and if the farmers left prices unchanged, the price level would be above equilibrium. How hard would it be to motivate those farmers to lower their prices and get the price level down?

Of course, if your model has people projecting past inflation into the future (as might be appropriate when there is no long run equilibrium price level, but rather a price level that just evolves depending on actual inflation,) then the equilibrium adjustments in the price level would cause people to expect that the rate of change in prices will persist. There will always be overshooting.

The problem will be even worse if inflation adjusts mechanically according to past unemployment rates. The change in the current level of output needed to get nominal GDP on target given the level of inflation that is already determined results in an unemployment rate that requires a change in the inflation rate the next period. Again, this requires constant overshooting.

But in reality, there is no need to adjust prices and wages “next period” based upon unemployment “this period.” That is a sensible strategy if the growth path of nominal expenditure has permanently changed. For example, nominal expenditure on output has shifted to a higher growth path. Inertia in price setting results in firms responding mostly by raising production in the first period. The unemployment falls. This provides a signal to firms that nominal expenditure is rising more rapidly, and so they raise prices more rapidly in the next period. This brings the growth path of prices closer to equilibrium with the new growth path for nominal GDP, and unemployment falls closer to its natural level.

But if lower unemployment this period doesn’t provide any reason to believe than nominal GDP on a higher growth path next period, then raising prices faster next period is not equilibrating. If, in fact, what happens to nominal GDP next period depends on its current value relative to its target, then next period’s price setting will not be influenced by this period’s inflation rate. McCallum explained the problems with the Ball paper years ago.

In my view, overshooting with painful consequences could well be the result of a shift to a new regime. However, the assumption that no one will ever learn, and will follow pricing and production practices that only make sense in a “who knows what will happen to nominal expenditure” regime is implausible. Nominal GDP growth path targeting is less complicated for firms trying to set prices than inflation targeting and much less difficult in a nominal expenditure is just a random walk regime.

By the way, if it turns out that firms really are willing to set prices at infinity even though they will sell zero output, or else, will give products away even though sales are at an infinite level, I will admit that targeting the growth path of nominal GDP is a mistake. I will probably retire from economics too.