Tom Holden on monetary policy
In recent years, I’ve become depressed by the state of research in macroeconomics. I find many new research papers to be almost unreadable. Perhaps this reflects the fact that my own work is increasingly outside the mainstream. Thus I was very pleasantly surprised to see a new paper by Tom Holden that embraces many of the themes that I have been emphasizing. Even better, the paper is extremely well written (unusual for a macro theory paper) and is forthcoming in the highly prestigious journal Econometrica.
Before discussing Holden’s paper, let me be clear that I’m not suggesting that he necessarily agrees with my overall view of macro. He uses more of a New Keynesian approach whereas I am a monetarist, and he advocates inflation targeting while I prefer NGDP targeting. But on a number of important points we end up in the same place, even if we arrive there by different routes.
Consider the following claims, which sound vaguely monetarist:
In this model, only monetary policy shocks affect inflation. Of course, if there is a nominal rigidity in the model, monetary shocks may have an impact on real variables. But as long as the central bank follows a rule like this, these real disruptions have no feedback to inflation. Causation runs from inflation to real variables, not the other way round. We can understand inflation without worrying about the rest of the economy . . .
This is consistent with causation only running from inflation to the output gap, not in the opposite direction. Likewise, Miranda-Agrippino & Ricco (2021) find that a contractionary monetary policy shock causes an immediate fall in the price level, while impacts on unemployment materialise more slowly. Again, this suggests that causation runs from inflation to unemployment, not the other way round.
In the traditional Keynesian model, causation runs from real shocks (more physical purchases and more hiring) to nominal outcomes (higher wages/prices.) Milton Friedman saw causation running from nominal shocks (money and inflation) to real effects (more jobs and output). Both views of causation are consistent with the correlations observed in Phillips Curve studies, but the monetarist interpretation tends to nudge people more toward monetary policy as the key stabilization tool.
Here is Holden’s policy proposal for stabilizing inflation:
Unlike with previous Taylor Rule proposals, Holden envisions deriving the real interest rate from inflation-indexed bonds, i.e. “TIPS”.
In the past, I’ve argued that economists focus too much on the public’s inflation expectations, and that the key to successful monetary policy is stabilizing the expectations of financial market participants. Here’s Holden:
The only expectations that matter are the expectations of participants in the markets for nominal and real bonds. It is much more reasonable to assume financial markets lead to prices consistent with rational expectations than to assume rationality of households more generally.
This is all music to my ears. Here’s another gem:
Real rate rules also have a second source of robustness: they do not require an aggregate Phillips curve to hold. The slope of the Phillips curve can have no impact on the dynamics of inflation. If a central bank is unconcerned with output, they do not even need to know if the Phillips curve holds, let alone its slope. Nor does it matter how firms form inflation expectations. The Fisher equation and the monetary rule pin down inflation, so while non-rational firm expectations could affect output fluctuations, they will not alter inflation dynamics.
I’ve also argued against using the Phillips Curve in monetary policy. I favor stabilizing market expectations of NGDP, while Holden is proposing the stabilization of market inflation expectations, but the underlying approach is the same—stabilize market expectations of a nominal macro goal variable. Don’t try to manipulate the Phillips Curve.
I’ve emphasized that any successful monetary policy regime leads to almost complete monetary offset of other demand side factors, such as tax cut-financed fiscal stimulus. Holden goes even further with monetary offset, as he is proposing an inflation target. So his proposed policy rule also offsets supply side influences on inflation, although he later argues (correctly) that policymakers may wish to adjust their target when there are supply shocks.
In my own work, I’ve strongly criticized the view that monetary policy works by changing interest rates, at least in the Keynesian sense of impacting the economy through changes in both nominal and real interest rates. I’ve created various thought experiments where prices are flexible and the effects of monetary policy on nominal aggregates cannot possibly derive from changes in real interest rates. Holden makes a similar claim:
An even more fundamental question of monetary economics is “how does monetary policy work?”. The traditional answer involves movements in nominal rates leading to movements in real rates, due to sticky prices. But this cannot be the transmission mechanism under flexible prices, as then real rates are exogenous. Nor too can it be the transmission mechanism under a real rate rule, as then real rate movements are irrelevant. In these cases, monetary policy works exclusively through the Fisher equation’s link between nominal rates and expected inflation. Since we will see that dynamics under a real rate rule are qualitatively so similar to dynamics under a traditional rule, it would be surprising if monetary policy worked by a fundamentally different channel under a traditional rule. Instead, this suggests that the main channel of monetary policy in New Keynesian models is the one also present even under flexible prices, via the Fisher equation. Rupert & Šustek (2019) draw the same conclusion based on the observation that contractionary (positive) monetary shocks can lower real rates in New Keynesian models with capital.
I have argued that a contractionary monetary shock actually lowered real interest rates in 2008, but it also lowered NGDP—creating a severe recession.
During the 1980s, a number of economists including Earl Thompson, Robert Hall, David Glasner, Robert Hetzel and myself proposed policies that would effectively target the financial market forecast of inflation or (in my case) NGDP growth. Holden suggests that his real rate rule is in that tradition:
Additionally, in older work, Hetzel (1990) proposes using the spread between nominal and real bonds to guide monetary policy, and Dowd (1994) proposes targeting the price of futures contracts on the price level. This has a similar flavour to a real rate rule, as these rules effectively use expected inflation as the instrument of monetary policy.
Forecast targeting has also been proposed by Hall & Mankiw (1994) and Svensson (1997), amongst others.
In the past, I’ve suggested that the Fed’s interest rate target should be adjusted daily, not every 6 weeks. Here’s Holden:
Note that while under conventional monetary policy, nominal interest rates are approximately constant between monetary policy committee meetings, this may not be the case here. . . . the central bank’s trading desk could have to continuously tweak the level of [interest rates] . . . While this is a departure from current operating procedures, there is no reason why holding [TIPS spreads] approximately constant should be any harder than holding [interest rates] approximately constant. This is thanks to the real-time observability of [real interest rates] via inflation-protected bonds.
I’ve argued that central banks should determine the strategy of monetary policy (i.e. whether to target prices or NGDP, and whether to target levels or growth rates), whereas market expectations should be used to actually implement the policy. Holden concludes his paper with a similar observation:
We have presented a design for the practical implementation of a real rate rule with a time-varying short-term inflation target. Under this proposal, central bank boards keep the crucial role of choosing the desired path of inflation. Only the technical decision of how to set rates to hit that path is delegated to the rule. The rule embeds no politically sensitive views about the slope of the Phillips curve or the costs of inflation. And the rule can be implemented using assets for which there is already a liquid market: either nominal and real long-maturity bonds, or inflation swaps.
In my recent book, I steered clear of the issue of “indeterminacy” (i.e. multiple possible equilibria), which is an issue where I don’t have expertise. Based on what I have read, however, it seems to be a bigger problem with interest rate targeting than with monetary regimes that stabilize a price, such as the gold standard or a fixed exchange rate regime. I suspect that indeterminacy is less of a problem with a real rate rule because TIPS spreads are analogous to a CPI futures contract, and hence stabilizing TIPS spreads is akin to targeting the price of a CPI futures contract. A gold standard avoids indeterminacy because gold prices are visible and controllable in real time. The same is true of CPI futures contract prices. If this is inaccurate, please correct me in the comment section.
Although I favor NGDP targeting, I believe Holden is wise to frame his proposal as an inflation-targeting regime. Unlike with NGDP expectations, we already have deep and liquid TIPS markets, and real world central banks have opted for inflation targeting over NGDP targeting. Framing the proposal as an inflation-targeting regime is the best way of moving real world policymakers toward the broader goal of targeting market expectations of the goal variable.
