Thursday, November 12, 2020

1933 lessons for today

Nov 11, Eric Leeper presented "Recovery of 1933" with Margaret  Jacobson and  Bruce Preston, at the Hoover "Road Ahead for Central Banks" series, and it was my pleasure to discuss it. This is a really important and insightful paper.  

Since Japan hit the zero bound more than 25 years ago, economists have been thinking about how to avoid deflation. The answer seems obvious -- "helicopter money," or "unbacked fiscal expansion." But this has proved remarkably hard to do. Jacobson, Leeper and Preston show us how the Roosevelt Administration managed a credible unbacked fiscal expansion, and it bears important lessons today. 

(Today we will not talk about whether raising inflation is a good idea. Central banks have wanted more inflation, so let's figure out how it could be done.) 

To think about these issues, let me write the basic government debt valuation equation, real value of debt equals present value of real primary surpluses.  

\[ \frac{B_{t-1}}{P_{t}} = E_{t} \sum_{j=0}^\infty \rho^j s_{t+j} = E_{t} \left( s_t + \rho \sum_{j=0}^\infty \rho^j s_{t+1+j} \right) = E_{t} \left( s_t + \rho \frac{B_{t}}{P_{t+1}} \right) \] 

Eric and I are fiscal theory fans, in which causality goes from right to left in this equation, but you do not have to take this view. The equation holds in all (almost) models, and we can think about the fiscal backing or coordination with monetary policy just as easily as the fiscal causes of inflation. 

An "unbacked fiscal expansion" means deficits, negative s, on the right hand side. That should push up the P on the left hand side, no?  But the equation points to a central verity. The stimulative effect of deficits does not depend on today's deficit alone. It depends on the entire expected stream of current and future deficits and surpluses. A deficit today, a negative \(s_t\), that comes with expected higher subsequent surpluses, higher \(s_{t+j}\), in such a way that the sum does not change, makes no difference at all.  And indeed most deficits do come with implicit or explicit promises of future surpluses to pay off the debt. As the right hand versions of the equations emphasize, when a deficit comes with an increased value of debt, we know that investors expect more surpluses in the future. How do you convince people that a deficit today does not come with a promise of future surpluses to repay the debt? That's the hard part of unbacked fiscal expansion. 

As Jacobson, Leeper and Preston point out, conventional Keynesian stimulus does exactly this sort of thing, which in this analysis is no stimulus at all. The Obama stimulus, like many, came with promises that yes, when it's over, we'll go back to surpluses and pay back the debt, all of the debt, even that coming from the stimulus payments. As Chris Sims pointed out in discussion, even the government of Japan has always offered reassurance that yes, it plans to pay off its large debts, for example introducing consumption taxes. Politicians really don't want to even say, "we are not paying off this debt, we're going to inflate it away." 

And for good reason. It might be easy to convince people that we're going down the Venezuelan path, defaulting on all debt, sparking a hyperinflation, and thereby establishing a reputation that precludes borrowing again for a long time.  But the government wants to inflate away just a little debt, while at the same time preserving its reputation that when things get back to normal, future borrowing will be accompanied by subsequent surpluses to pay back that debt, not by further inflations or defaults. How do you convince people that you're just swiping the tip jar just this once, but that doesn't mean you're embarking on a life of crime? 

Our governments have earned a reputation for, so far, mostly repaying rather than inflating away debts. When they issue treasury bonds to finance deficits, negative \(s_t\) the value of subsequent debt goes up, second terms in the right two equalities, the governments raise revenue from the debt sales, and they don't regularly inflate away outstanding debt. Overcoming this expectation, just a little bit, and just this once,  is really hard. People know to discount politician's speeches and promises! 

Helicopter money is an unbacked fiscal expansion. When interest rates are zero or reserves pay market interest, money and bonds are the same thing, so dropping money or bonds from helicopters is the same thing. And they are fiscal policy. The Federal Reserve is not allowed to drop money from helicopters. The Fed must always buy some asset in exchange for money. Dropping money from helicopters, is the same thing as writing checks to voters. It is fiscal policy, it counts as the deficit, and the Congress and Treasury must do it.

Even a helicopter drop, if followed with an announcement "taxes will be going up tomorrow to soak up all this money" would produce no inflation. The future s terms matter for helicopter drops too. Helicopters are a lovely fiscal device, to try to convince people that this debt issue will not be repaid, though when the Treasury goes back to auctioning bonds, those debt issues will be repaid.  The  paper is about how to construct a similar commitment to not repay debts in order to produce a little inflation, and preserve our ability to borrow in case... well in case 1939 and 1941 roll around. 

To the point. What does 1933 teach us?  As you know, the Roosevelt Administration faced a large deflation, much worse than the mild few percent we have seen. They felt this was contributing to the depression via debt-deflation -- many people had taken out loans that they could not repay at the much lower price level. They wanted inflation, a lot of it, now. 

"One fact figured prominently in his [Roosevelt's'] thinking: the precipitous decline in overall price bankrupted the famers and homeowners... FDR felt that the key to economic recovery lay in retiring overall prices to their 1920s levels...the problem was that in the 1920s the price level was 60 perfect above the long-run average to which it had to revert to maintain gold convertibility at the parity that prevailed over the previous century.'' 

Their problem was doubly acute. The US was on the gold standard. Now, the gold standard is essentially a fiscal promise. Should paper dollars start to lose value relative to gold, people come get gold. The government must raise taxes or borrow against future taxes to get the gold. The gold standard is a way of committing that the \(s_t\) on the right hand side of my equation will be there to pay off the bonds \(B\) at the desired gold value. It is a great recommitment against printing up too much money. 

The gold standard has a big flaw though. It produces a stable price level in terms of gold, and commits the government to repaying debt at the gold value, but it does nothing to stop the situation that both gold and money become more or less valuable relative to goods and services, which is what we care about in the end. Thus there can be substantial inflation and deflation under the gold standard, 

The gold standard as a fiscal commitment has a worse flaw. Back to our equation. P represents the price level in terms of goods and services. Now, if a deflation happens, under the gold standard the government is obliged to raise taxes and cut spending -- to raise surpluses -- in response to a deflation.  P on the left hand side goes down, s on the right hand side must go up.  The fiscal policy is "active" and determines the price level in terms of gold, stopping the government from printing money and inflating relative to gold. But the fiscal policy is "passive" and must accommodate changes in the price level when both gold and money become more valuable.  It forces an unbacked fiscal contraction to validate deflation. 

Roosevelt's problem was worse. If deflation had just broken out of nowhere, then the gold standard would still fix the long-run price level. But the price level was still above its long run average. There had been a lot of inflation relative to gold in WWI. A lot of debt had been taken out in the 1920s at higher price levels. The big 1933 deflation was really only the beginning of a return to trend of the price of gold and dollars relative to goods and services. So there was little chance of the price level recovering under the gold standard. But Roosevelt wanted to return to a permanently higher price level. 

Roosevelt took office in March 1933. The deflation stopped immediately and the price level rebounded. 

What were the key policies? First, devaluation, "sharply reduced the gold content of the dollar.'' Second, "'emergency' fiscal deficits financed by new issues of nominal Treasury bonds....the modifier `emergency' also communicated the temporary and state-contingent features of the fiscal program." Third, rhetoric. "Roosevelt made recovery the policy priority; higher, for example, than the last century's fiscal orthodoxy..'' The New Deal was indeed radical, changing forever the nation's conception of the Federal role in economic policy. Much of it was disastrous - - the NRA, 1937 tax increase, "war on capital," and so on. But for the point here, it is reasonable for people to believe a really sharp break in economic policy had occurred. 

The first is obvious. Devaluation is a way of credibly committing that bonds will be repaid in dollars that are worth less. A credible, permanent devaluation will cause inflation, moving the future price level trend up.  In fiscal theory, a promise of future default triggers inflation today -- people try to sell bonds today since they will be worth less tomorrow. A devaluation is a default under the gold standard. 

The emergency budget device is the thing I learned most from this paper. By saying "emergency," and by formally separating the two budgets, Roosevelt was able to signal and commit that this part of the budget would be inflated away, but that part of the budget, even if in deficit, will be repaid. 

Here are some heuristic equations to illustrate the idea. The government follows a fiscal rule, which we might write \[ s_t = \theta_y (y_t - y^\ast) + \theta_p (p_t - p^\ast) +\alpha \frac{B_{t-1}}{P^\ast_t} \] The first term represents the emergency budget. The government will run deficits as long as output is depressed. That's the "state-contingent" art of the paper's analysis. Similarly, the government will run deficits so long as the price level is below the price level target, communicated here by a change in gold content of the dollar. I add a third term, to suggest repayment. Here I wrote that the government will repay debts, but only at the price level target. It will ignore deflations. This is a bit heuristic. My last paper, here writes a more formal version of this idea. You might write instead a commitment to repay deficits from the ordinary budget but to ignore deficits from the emergency budget. Writing this down in a more formal way is a really important task. The paper includes such a model (p. 30), but it's more complex than I can summarize in a blog post. I would also like to see a simple example model that really clarifies the idea as well as the paper's more realistic estimable but inherently more complex model. 

You intuit though how such a fiscal rule would be very useful to stem deflation, and more generally to help to control the price level. I call it a "fiscal Taylor rule," with apologies to John Taylor. In a modern economy that is not tethered to the gold standard, one might view such a rule as an implementation of a price level target, or modify it to react to inflation rather than the price level. 

Lessons for today

This paper has many important lessons for today. Obviously it suggests some fiscal rules and institutions that might help governments to implement modest unbacked fiscal expansions if they feel deflation is a problem, while maintaining their reputations for debt repayment of normal borrowing. The emergency budget and fiscal Taylor rule are a great start. 

When 2008 came, many economists thought, oh no, here comes 1932 again -- we're at the zero bound, monetary policy is useless, here comes a deflation spiral, just as our ISLM models predict. It did not happen. 2008 and beyond was severe, but we did not see a 30% decline in the price level. Even Europe and Japan never did worse than a percent or so deflation. Why is that? 

The fiscal analysis answers the question beautifully. Because our governments did not and were not expected to passively validate deflations, as gold standard governments were. If a 1932 30% price level cut breaks out on the left side of our equation, the "passive" fiscal policy in standard models says that just as in 1932 our governments will be committed to massive fiscal austerity, raising surpluses 30% to pay a windfall to bondholders. (To Wall Street fat cats and Chinese central bankers, while America is suffering, as our politicians would have put it.) There is no way any of our governments are going to do that, and nobody expects it.  Indeed, the government did not do it in 1933, when there was an explicit promise to do so. The concept of passive fiscal policy should have died in 2008.   

Before we even get to unbacked fiscal expansion let us recognize how good the current system is. Our governments are credibly committed to fiscal inaction -- that they will not respond to a big deflation by fiscal austerity. That's why we didn't have 1933 to contend with in the first place. And that's why I think the demand for unbacked fiscal expansion is overblown. We do not have to fight against a deflation spiral ready to break out at any moment. The worst that can happen is a few years of gentle 1% or so deflation -- live the Friedman rule. 

The second lesson for our time, a bit more speculative: Why has inflation been so surprisingly low, and declining, despite fiscal ... profilgacy? ? Well, generalize our equation to include variation in the real interest rate, \[ \frac{B_{t-1}}{P_{t}} = E_{t} \sum_{j=0}^\infty \frac{1}{R^j} s_{t+j}.\] A decline in the real interest rate makes the stream of surpluses more valuable. (Equivalently, it lowers interest costs of the debt, so is equivalent to extra surplus.) That's a deflationary force. So the slow and steady decline in real interest rates, for whatever reason, seems consistent with the slow and steady decline in inflation. 

In sum, this is a great paper with many eye opening lessons for today. 

(BTW, I quickly reviewed the paper at 2016 next steps for FTPL conference. It keeps getting better.)  


  1. Your last part once again suggests that central bankers may have got it all backwards: if they want higher inflation, they should RAISE interest rates!

    However, one part that I always find odd is that central banks obsess over "low inflation" (CPI) that is actually only a few tenths of one percent below their target (a rounding error). While at the same time the actual REAL LIFE value of money shows massive inflation, with for example house prices out of control and in the US also things like healthcare costs. Perhaps flatscreen TVs are getting cheaper but in the things we actually NEED we see massive inflation.

    So instead of obsessing over a few tenths of one percent miss in "inflation" target and throwing everything but the kitchen sink at it to raise this "inflation", why not challenge the way we currently measure "inflation"? Because it seems completely wrong and arbitrary to me. The poorest 50% of the population who get priced out of everything despite many years of "growth" will be grateful.

  2. I have Power to coin vs Power to tax.
    Who wants to replace those remits back to Treasury with a one point federal sales tax? But, who wants Treasury stepping on the power to tax? The two will zero cross.

  3. Don't forget about sterilization by the Fed, which prevented inflation.

  4. And the difference between the approach taken by FDR and the approach advocated by proponents of MMT is...?

    The advocates of MMT would appear to have a point. "If FDR can do it," they say, "then why can't we?"

    It may come as a surprise to some, but the average bond investor is an insurance company, a pension fund, or a little old lady clipping interest coupons. The insurance company and the pension fund payout to, [surprise!] households in the form of insurance payouts on life and casualty policies, while pension funds payout retirement benefits and survivor's benefits over a lifetime. Not every investor in bonds is a Wall Street Journal, is a "Wall Street fat cat." One might then wonder how it would be socially beneficial (i.e., raise welfare) to de-fund the insurance industry, the pension industry, or the pensioned-off widower?

    1. Yep, I seem to remember an article from John that described MMT as "Magical Monetary Theory".

      "One might then wonder how it would be socially beneficial (i.e., raise welfare) to de-fund the insurance industry, the pension industry, or the pensioned-off widower?"

      See the merger between insurance and investment banking (Gramm-Leach-Bliley) to understand what happened to the "bond market vigilantes".

  5. I did a lot of research on this question back in the 1990s. There's pretty strong empirical evidence that the inflation that occurred after March 1933 was mostly caused by dollar devaluation, with fiscal policy only a minor factor. That's not to say devaluation has no fiscal impact--it is a sort of default.

    1. But where does the dollar devaluation come from? This is entirely consistent with the fiscal story. The exchange rate is the least sticky price. It falls first, then commodity prices, then stickier prices.

  6. Devaluation was simply an administrative decision, where fiscal policy played no role. FDR changed the definition of the dollar (with the stroke of a pen) from 1/20.67 oz of gold to 1/35 ounce of gold.

    You don't need fiscal policy to change a definition, you just need to commit to exchange dollars for gold at the new price ($35/oz.) on demand.


Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.