Arnold Kling  

Shorter Scott Sumner

Bruce Charlton on IQ, Educatio... The Financial Crisis: A Matri...

Bryan points to a Scott Sumner's monetarist credo.

Think of it as a simplification of the Taylor rule. The Sumner Rule is, if the consensus forecast of nominal GDP growth is below 5 percent, then loosen up. Presumably, if the consensus forecast is above, say, 7 percent, then tighten up.

Sumner writes,

When the growth rate of nominal GDP falls sharply there is always a severe recession.

Here is the latest data (subject to revision) for percent changes at an annual rate:

QuarterNominal GDPReal GDP
2008 q24.122.83
2008 q33.35-0.51
2008 q4-5.77-6.34
2009 q1-2.87-5.49

In the second quarter, we are a bit below the Sumner rule for nominal GDP, but basically we are doing OK. In the third quarter, we drop to 1.65 percentage points below the Sumner Rule, but real GDP growth plunges by twice as much. Then both nominal and real GDP collapse further. Sumner's story for the third quarter would be that people saw the drop in nominal GDP growth coming, and so real GDP growth fell.

Suppose the Fed had adopted a policy of all-out expansion to try to maintain nominal GDP growth of 5 percent. Also, for fun, suppose no bailouts and no stimulus--that is, no other policies to fight the crisis. What would have happened?

Without bailouts, presumably we get more bank failures--and Freddie and Fannie go under. The Fed meanwhile buys up short-term government debt until the interest rate hits zero, then goes to work on intermediate and long-term government debt, and if those rates hit zero it starts in on mortgage securities.

My guess is that in the short run, nominal GDP and real GDP fall anyway. I just don't think that the private economy can re-allocate employment and consumer demand that quickly. Without the bailouts, there might be more panic and job loss in the financial sector, but less panic and an earlier return to profitability in the nonfinancial sector. Overall, my guess is that a durable recovery would have started sooner, but I do not think that the Sumner Rule could have prevented a short, sharp recession.

An interesting thought-experiment, though.

Comments and Sharing

COMMENTS (10 to date)
Ironman writes:

In the second quarter, we are a bit below the Sumner rule for nominal GDP, but basically we are doing OK. In the third quarter, we drop to 1.65 percentage points below the Sumner Rule, but real GDP growth plunges by twice as much.

Following the "Sumner Rule" though, wouldn't you still have had a significant policy response from the Fed much sooner? If you go back to 2007, here's how the quarterly nominal GDP annualized growth rates stack up:

2007Q2 ... +6.88%
2007Q3 ... +6.35%
2007Q4 ... +2.33%
2008Q1 ... +3.45%
2008Q2 ... +4.12%
2008Q3 ... +3.35%
2008Q4 ... -5.77%
2009Q1 ... -2.87%

As such, you would have had a significant policy response beginning as early as the first quarter of 2008. With that, do you get the same sharp recession? (Assuming of course for the terms of the thought experiment that the oil shock of the summer of 2008 doesn't take place....)

Bill Woolsey writes:


Your forgot the minus sign on real growth in the first quarter 2009.

Scott doesn't disagree what if some kind of disruption requires a realocation of resources, then real output will shink and prices rise faster.

What he disputes is your claim that consumption cannot shift so fast so that nominal income will fall anyway.

Part of his argument is that if nominal income in the 3rd quarter 2009 is expected to be 5% higher than it was in the third quarter of 2008, it just won't drop much during the 4th quarter of 2008, or the first and second quarter of 2009.


Scott isn't proposing a feedback rule. So, the response on policy would have actually been during the 4th quarter of 2008, before the actual figures came in at the end of 2008.

Clare Zempel writes:

For much of the Greenspan era, it was as if the Fed targeted Nominal GDP growth in the 4-6% range. The Fed tended to tighten when Nominal GDP's four-year growth rate rose above 6% and seemed likely to remain strong. It eased when it slipped beneath 4% and seemed likely to remain weak. The implications were that the Fed assumed that Real GDP's sustainable trend was around 3% and that it targeted broad inflation to average around 2%.

Radford Neal writes:

I don't know whether or not maintaining 5% nominal GDP growth is a good idea or not. But I do think that it's possible, whereas you seem to be arguing that it's not. Surely with sufficient actions and sufficient rhetorical committment to future actions, the central bank can produce enough inflation to achieve this, even in the short term. The only argument otherwise would be that drastic monetary expansion, and the threat of more of that, might reduce real GDP (due to general economic panic) by even more than it increases inflation, but at some point, the inflation is going to dominate, since people are still going engage in trade in order to eat. Even Zimbabwe doesn't have zero GDP...

Kevin Dick writes:

To expand on Ironman's point, your analysis assumes the _vector_ going into the recession would be the same. This is certainly true if we started using the Sumner rule in 2Q08. However, if we'd been using the Sumner rule all along, the vector would have been adjusted by _both_ an earlier policy response _and_ different expectations of the response.

I don't know what the effect would be but I'd be curious to hear how you think different expectations would have shaped the adjustment by the real economy. Would people have started adjusting earlier? Would they make more extreme adjustments? Would they reallocate activity differently?

fundamentalist writes:

If Sumner is right, then why hasn't the Feds' loose monetary policies since 2007 caused an increase in NGDP?

Scott Sumner writes:

Thanks for the post Arnold, I think it is helpful to think of what would happen if the Fed pegged a 12-month NGDP futures contract at a price 5% above current NGDP. In that case there would have never been a sharp drop in NGDP growth expectations in late 2008. The problem with many counterfactuals is that they assume my 5% NGDP growth target, but then plug in other numbers that are totally inconsistent with that NGDP number. For instance, if NGDP growth was expected to be 5%, we would never have been in a liquidity trap, as very low nominal interest rates reflect low expected NGDP growth, not "easy money." Also, the banking crisis would have been far milder. Only the first, relatively mild part of the banking crises in late 2007 and early 2008 was due to foolish loans. The much bigger part of the crisis that occurred in late 2008 occurred precisely because sharply falling NGDP dramatically reduced asset values around the world. So it makes no sense to try to imagine what would have happened with a forward-looking NGDP rule plus a liquidity trap and a severe fiancial crisis, because with 5% NGDP targeting there would have been no liquidity trap and the financial crisis would have been much milder. In addition, industrial production in Asia and Europe might not have fallen sharply, and we would have missed the second leg down of the US housing crisis, which was even more severe than the original subprime housing crisis.

BTW, I favor something called "level targeting" which is a 5% target path or trajectory. Thus we could probably benefit from 6% NGDP growth over the next year, and 5% thereafter, as we have already fallen at least 1% behind the path I proposed in February.

fundamentalist writes:

Scott, are there any historical periods in which your theory was applied?

Arnold Kling writes:

What does it mean to take a position in a futures contract on nominal GDP?

For example, the next data on GDP to come out will be for the second quarter. The Fed cannot manipulate that quarter--it's all in the past.

Can the Fed affect the third quarter? Probably not. Most consumers and businesses are not going to all of a sudden change their plans in the third quarter. If nominal GDP goes up by 5 percent, it will be due to luck in relation to the inventory cycle or somesuch.

Can the Fed affect the next four quarters? Now we're talking about something more plausible. However, it is possible that if the Fed hits a target of 5 percent nominal GDP, it will mostly turn out to be inflation in sectors already close to full employment, rather than shifts of workers and capital out of banking, autos, and construction into those sectors.

Scott believes in a world in which economic actors are making major choices based on expectations for nominal GDP. I believe in a world in which the individual consumer is forming expectations about his own income for next year, and the individual business leader is forming expectations about demand for her own company's offerings for next year. To heck with somebody's nominal GDP forecast.

Bill Woolsey writes:


You should read some of Sumner's explanations of the proposal, he is promoting a targeting nominal GDP one year in the future.

To take a position on an index futures contract on nominal GDP is to buy one from the Fed or sell one too the Fed.

When the contracts are settled, if nominal GDP is above target, the Fed pays those who bought and collects from those who sold. If nominal GDP is below taget, the Fed pays those who sold and collects from those who bought.

The Fed makes ordinary open market operations in bonds now based upon the trades that other people make regarding nominal income index futures.

The Fed adjusts base money today with ordinary open market operations, so that other people buy and sell matching amounts of futures. The Fed is hedged and, in nominal income turns out to be away from target, it tranfers money between those who have sold and those who have purchased.

I certainly agree that people make decisions based upon what they expect to happen to themselves or their firms. But expecations about the aggregates do impact this.

Recessions are lower income. Lower income lowers the demand for normal goods. The derived demand for intermediate goods will fall too.

As for the "full employment sectors" having inflation. Exactly. There will be an increase in demand, higher prices, and higher profits in sectors where demand is increasing. There will be lower output, losses, and lower prices in sectors where demand is falling.

If output can shink faster in sectors where demand is shrinking than it can rise in sectors where it is rising, and prices don't in sectors faster than they rise in expanding sectors, then the inflation rate rises and real output falls.

But we have sectors in the economy with high prices, high profits, and a motivation to expand production as fast as is feasible.

Along with other sectors with excess capacity and downward pressure on prices.

As resources are redeployed, prices fall (or rise more slowly) in the sectors where demand as increased. Profits fall again. Equilibrium is reestablished.

I think this is a good macroeconomic environment for relative price adjustments. If there is a substitution of less risky for more risky production processes.. fine. If there is a shift from investment to consumption, fine. If people want to work less (or spend more time cleaning house) fine.

Some parts of the economy expand and others shrink.

Prices rise in sectors that need to expand. Nominal profits rise in sectors that need to expand.

And there should always be a rough balance where expanding sectors will outstrip weak sectors based upon growth in productive capacity.

Now, if people want to work less, or there is some other reduction in productive capacity, then prices rise. That is the way nominal income targetting works.

In reality, nominal income today (or at the end of last quarter 1 2009 is below where it was at the end of quarter 1 2008. If your preferred system works just as well, then despite the falling nominal income, there should be sectors of the economy expanding as fast as is feasible. Where are these sectors?

I see figures that showed drops in nearly every type of consumption and investment. It looks to me like production of about everything is dropping. Are you going to stick with the view that everyone is working at home? It isn't worth working for money, because the goal is saving, and there is no way to save without too much risk. So, paint the house?

You are implicitly assuming that all prices and wages always adjust so that real expenditures is always equal to real productive capacity and that any decrease in real output must be due to some kind of drop in capacity. Yes, you have already said that there is a shift going on and resources are hard to shift. Where are they shifting too? Where are these sectors that require more resources?

Comments for this entry have been closed
Return to top