Exchange between Arrow and Davidson on debt

Dear Professor Davidson,

I must respectfully but strongly disagree with the implications of the statement by you, Lord Skidelsky, and my friend, Jamie Galbraith. The existence of a large public debt is a burden, especially in that it will inhibit deficit spending in future recessions. Deficits require someone to buy the newly issued securities; people already holding a large amount of government debt will demand higher and higher rates of return to hold more.  There was an argument about the “burden of the debt,” in the 1960s; it was summed up in an excellent paper by Franco Modigliani, who, however, did not go into the implications for future counter-cyclical policy.  Without going into details here, the point emphasized by Modigliani was the displacement of private investment. His model presupposed the absence of the Ricardo effect; if there were a Ricardo effect (prediction of future tax burdens due to debt issuance), then Keynesian policies could not possibly be effective, since people would increase saving to compensate for the deficit. Hence, the same argument that deficit financing is stimulating implies that public debt can be a burden. 

I understand your concerns that the argument can be used to decrease social security and government medical support. It can equally well be used to justify increased taxes to pay for them. 

In any case, as the proportion of population over 65 increases, it is clear that in fact something does have to be done. There is one obvious step: increase the age of retirement. As people are becoming healthier at any given age, this is a reasonable (to me, obvious) policy. It is a kind of decrease in retirement benefits (including health) but is compensated for by an increase in national income through an increase in labor force participation.

You have cited the historical record that subsequent prosperity has kept the debt manageable. In the case of the Clinton administration, there was a deliberate policy (including tax increases) to achieve that.  The continued growth in desirable entitlements means that the past is by no means a reliable predictor of the future.  We can see what has happened to Japan.

I would appreciate your forwarding this message to Lord Skidelsky and James Galbraith. 

Sincerely yours,

Kenneth J. Arrow

——-

---

 ———————————————————————————————

Dear Professor Arrow:

Thank you for your thoughtful response –even though you strongly disagree with the statement by Jamie Galbraith, Robert Skidelsky and myself.

Perhaps I can suggest some reasons that I think may encourage you to rethink  your feeling of disagreement.  Please note that what follows is simply my own reaction to your comments – and does  not reflect how Jamie Galbraith or Robert Skidelsky might respond to your note. After al, the three of independently decided not to sign the Evans letter.  [I shall forward a copy of your letter to them as per your request.]

The response to Sir Evans letter, in my case, derives from what I believe is a  proper understanding and application of Keynes’s analytical framework to our current Great Recession.  The Keynes framework that I am referring to is NOT  the neoclassical synthesis “Keynesianism” [expounded by Paul Samuelson, and accepted in the 1950s by almost all economists as a proper representation of Keynes].

            In my book THE KEYNES SOLUTION: THE PATH TO GLOBAL ECONOMIC PROSPERITY, I quote Paul Samuelson stating that on reading The General Theory , he found Keynes’s paradigm “unpalatable” and not comprehensible. Samuelson when on to state: “I was content to assume that there was enough rigidity in relative prices and wages to make the Keynesian alternative to Walras operative……”.  Yet in chapter 19 entitled “Changes in Money Wages” Keynes specifically states (on page 257) that the classical system assumed “the fluidity of money-wages” for a self adjusting mechanism and “when there is rigidity, to lay on this rigidity the blame for maladjustment … My difference from this theory is primarily a difference of analysis”.  So if Samuelson had read this page, he could not assert that Keynes was a slow adjusting Walrasian system with rigidity of wages. Keynes’s analysis does not depend on the rigidity or stickiness of wages.. 

For Keynes the fault lay in the operation of financial markets and the need for liquidity in a money contracting economic system . By liquidity I mean having the ability to meet money contractual obligations when they come do!  This liquidity concept is a meaningless concept in any Walrasian system – as you know quite well. Yet in Arrow and Hahn (p. 356-7) you noted “The terms in which contracts are made matter. In particular , if money is the good in which contracts are made, then the prices of goods in terms of money are of specific importance. This is not the case if we consider an economy without a past or future….”. This is part of the Keynes system that I think is relevant.

Remember Keynes on page 16 of the GT, argued that classical economists were like Euclidean geometers in a non-Euclidean world… and what was required was to throw out the equivalent of the axiom of parallels from the classical economic system to work out a non-Euclidean geometry for the non-Euclidean world of experience.

What axioms did Keynes throw out?  In his article in the Spiethoff festschrift, Keynes specifically stated that the neutral money axiom had to be rejected in both the short run and the long run for the real world. 

In Keynes’s emphasis on uncertainty (rather than probabilistic risk) regarding future outcomes of current decisions, Keynes was throwing out what I have labeled the ergodic axiom associated with ergodic stochastic processes (the ordering axiom plays the same role in deterministic models).   Keynes never knew the terms ergodic vs. nonergodic processes since this stochastic theory was developed in 1935 by the Moscow School of Probability and did not find it way to the Western Europe and America until after 1945.  But Keynes’s criticism of “Mr. Tinbergen’s method” was that economic data was not homogeneous over time.  And nonstationarity is a sufficient but not a necessary condition for a nonergodic process. Out goes the ergodic axiom. [By the way after significant personal discussions between John Hicks and myself, Hicks renounced the ISLM as a depiction of Keynes’s analysis and stated he wished he had labeled his view of the micro foundations of macroeconomics as a nonergodic view!]  

Finally in Chapter 17 entitled “The Essential Properties of Interest and Money” Keynes identified two elasticity essential properties associated with money and all other liquid assets, namely their elasticity of productivity is zero, i.e., money (and liquid assets) do not grow on trees and therefore private sector entrepreneurs can not hire workers to harvest money trees if the demand for other goods decline and the demand for money and liquidity rises. [See page 235 of the GT paragraph starting “Unemployment develops… because people want the moon, etc.]

The second essential  property is that the elasticity of substitution between all liquid assets and reproducible goods is zero.  Thus the gross substitution axiom is not universally applicable..  And as I believe Arrow and Hahn in their 1971 book stated if the gross substitution axiom was not universally applicable then all existence proofs were jeopardized.

Now how does all this respond to your Ricardo effect argument and also to Franco’s burden of debt argument. Both these arguments rely on some version of intertemporal optimization by decision makers today in a world where the future is known or at least actuarially certain..

First regarding the Ricardo effect. My attached article about making “Dollars and Sense out of the US Government Debt” implicitly suggests that the post world war 2 generation did not worry about a Ricardo effect. .After all when , as I point out in this attached article, the USA came out of world war 2 with a debt that equaled 120 of GNP at the time, I and my young cohorts at the time never worried about paying off the debt (did you at the time?) — and the evidence proved us correct.  At 80 years of age in 2010, I have never in my adult lifetime seen any of my cohorts savings just to pay off the national debt.. We were happy to inherit the “burden” of a prosperous economy – even if the government debt exceeded the GNP in 1945. In other words there was no optimal decision to allocate savings in order to pay off the debt when it matured – or any time later!

  Also note that the Republicans of the 1950s did not worry about paying off the debt as the  Eisenhower launched the largest peacetime public works program – the interstate highway system , while simultaneous maintaining probably the largest entitlement program at the time – the GI bill providing free education to all the  million of veterans if they wanted it. As a recipient of that entitlement program , I will be eternally grateful. (And mortgages with no down payments for housing was also part of the GI bill entitlements.)

Secondly you raise the issue regarding private sector bond holders demanding higher rates of return if we increase the outstanding debt.  I think Franco was just wrong and we do have some empirical  evidence to prove this. Just think of all the debt created in the last 30 years –since Reagan’s large deficits — and tell me if the rate of return has gone up or down since the 1970s?

At best I think your argument about these bond holders are what I may call the threat of  “bond market vigilantees” . Do you think these vigilantees can, and should, control what the US government does in terms of rates of return, liquidity, entitlements for senior citizens, job creation, etc. when there are vast amounts of unemployed resources available.  Just think of what Ben Bernanke has done for both the government bond rate and the mortgage rate – despite the housing boom of the last few years  – and the many bad mortgages out there Why has not the vigilantes forced the government bond rate up and the mortgage rate sky high?.  

Ultimately , the question of real support for Social Security ,Medicare, etc depends, in large part, on the productivity of labor and the labor resources available in the medical field  in the future – which no one today  knows with any degree of probability much less precision  All the modelers  “know” is that if past cost trends continue (ergodically) into the future, given the demographic trends, and given estimates of payroll taxes collected on some ergodic econometric forecasting model  , cash outflows from the “Social Security fund” will start to exceed cash inflows.

  In 2001 I had a debate with Tennessee Senators Bill Frist and Fred Thomson regarding privatizing social security. During our debate the Senators raised the issue that in x years there would be some many more elderly per employed worker than currently so that we (the employed population) could not afford to support the elderly in their comfortable social security way. Consequently the Senators argued we had to immediately raise the age of retirement to 70 so as to keep more members of the population working and paying into the fund and therefore there would be fewer retired senior citizens drawing down the fund each year in the future.

I pointed out to the Senators that if this debate was taking place 100 years earlier (in 1901), I could note that for every 20 mouths to feed in the USA at the time, it required over fifteen workers in agriculture and food distribution to provide enough food.  Suppose in 1902 I then told them that 100 years in the future (in 2001), there would be less than 4 agricultural and food distribution workers for every 20 American mouths to feed.  Would they say Americans would be starving to death in 2002 ? [ Bob Eisner , at the time, had a number of articles pointing to the fallacy of the cry that Social Security was doomed to bankruptcy if something was not done to prevent future deficits in the system.]

In fact near the end of your letter you state “”…the past is by no means a reliable predictor of the future”. So apparently you agree that the ergodic axiom is not applicable to our real world economy.  But if the ergodic axiom is not relevant, then what policy implications can we draw from any variant of an Arrow-Debreu-Walrasian intertemporal spot and forward market model –covering the indefinite future?

I hope you do not think I am being to forceful in putting forward my argument but I truly believe that many of our professional colleagues have mistakenly accepted the idea that the federal government’s debt is always a burden– and not , at less than full employment, a blessing to help us get back to prosperity and a civilized society.

You may want to read my 2009 book THE KEYNES SOLUTION: THE PATH TO GLOBAL ECONOMIC PROSPERITY to get a more detailed explanation of the points I am trying to make in this response to your letter.

  Paul Davidson