Why the Keynesian Multiplier is Negative in Peacetime

In the Depression of 1920, estimated GNP plunged 24% from $91.5B in 1920 to $69.6B in 1921. Unemployment jumped from 5% to 12%. The price deflator from 1920 to 1921 was the highest in 120 year history of time series, between 13% and 18% depending whose estimate. For contrast, the next highest was 11.5% in 1931-32.

New President Harding responded by slashing Federal government spending in half over two years, from $6.3B in 1920 to $5B in 1921 and $3.2B in 1922, vetoing various spending bills.

Result: GNP rebounded from $69.6B 1921 to $74.1B in 1922, unemployment dropping back to 6.7%, Eighteen months after Harding's inauguration, the twenties were already roaring. Unemployment continued to decline, reaching 1.8% in 1926. (Hat tips: Jim Powell Meltingpotproject.com).

This example seems to directly refute Krugman-style Keynesianism, which makes the straightforward prediction that cutting government spending $3B in a deflationary environment should decrease an already lacking demand, and cause a further drop in GNP of maybe $4.5B (to use the 1.5 multiplier stipulated in Obama's economists' analysis) -- the negative of what was actually observed. So its worth asking how the Keynesian theory might be confused.

Here's one possibility*. The reason demand is depressed right now is people feel poor, and unconfident about the future, so they are sitting on their money. But when Obama and Congress proposed and passed the trillion dollar stimulus, I personally felt much poorer. And so I am going to ratchet back my spending, yet again. I also felt less confident about the future of the United States, and thus less prone to invest in that future. I propose to the Demand side theorists that my response, and those of many who feel the same, will overwhelm any possible positive effect the stimulus might have on demand.

My belief that I am made poorer and less confident is quite rational (not just some matter of animal spirits). For one thing, its simply factually correct that I am poorer already, because the financial markets have run down, seemingly in response to the stimulus. Also, this stimulus seems destined to create extremely little wealth, to ratchet up the cost of government in perpetuity, to thus ratchet up my taxes and those of any business I might invest in, to ratchet up interest rates, and even to ratchet up the probability of hyperinflation.

Keynesian "theory" is based on assumptions that behavior (except of the individuals paid by government) remains constant, investment remains constant, prices remain constant, interest rates remain constant, etc. Clearly this is nonsense. The market responds instantly, perceiving that America is worth less with a much bigger government and huge future tax demands. Individuals' demand responds both to the market fall, and to their own perceptions of future higher taxes. Demand and investment are being crushed by the stimulus much more than any possible direct positive benefits. The "multiplier" is negative, maybe -2.

Conversely, if the Obama administration were to turn around, repeal the stimulus, and then pass a different bill which rather then spending $800 Billion, cut an additional $800 Billion, I would personally feel much richer and more hopeful. I bet that peoples' relief and feeling of being wealthier and more confident about our leadership and about the future, would ratchet up demand way beyond any lost through lessened government spending, and that the markets would stage a rally for the ages.

Anyway, my theory explains the current malaise as well as the Depression of 1920, the recovery in 1982 when Reagan cut spending 5%, and the various many failures of Keynesian stimulus such as 1929-39, the 1970's, and Japan in the 1990s. It even explains the (one! and even then, arguable** ) success of Keynesianism in WW2, because then consumption wasn't an option anyway, and people had more to worry about than government spending, as well as patriotic motivations. Government spending during WW2 was credibly temporary, and didn't have the negative effect on confidence it has today, just the contrary.

* Of course, numerous supply side reasons why the Keynesian picture is mistaken are widely appreciated. My point, here is to give a demand side picture so that the opponents of supply side come to grips with it.
From a supply side picture, for example: the stimulus will have to be paid for, requiring first borrowing (displacing private investment or consumption) and then requiring taxes to pay the debts. The taxes will impose dead weight cost: tax production and you get less of it, so that in addition to the amount the tax takes, some dead weight is never produced, and this dead weight may be as high as $2 for every $1 the taxes raise. (The 3 multiplier found by Romer and Romer for tax rate decreases also seems consistent with this: the 2 dead weight plus the 1 in tax itself.) The fact that the taxes are pushed into the future-- where they must be raised with interest, and in taxes yet to be determined -- increases uncertainty and if anything should exacerbate the effect on a rational investor. This picture suggests the multiplier of the current stimulus may be roughly -2, or maybe even more negative. If it is -2, the stimulus would take 4% off GDP per year (superimposed, of course, on whatever recovery might occur in the absence of the stimulus.)

** Barro found a multiplier of 0.8 for World War 2 spending, if all GDP growth was attributed to the spending (and thus less if some were attributed to normal growth or recovery processes). Cole and Ohanian found that the lack of normal recovery during the pre-war New Deal could be attributed entirely to artificial inflation of wages under first the NIRA and then the NLRA, and that this inflation vanished under wartime restrictions, and thus tend to attribute all wartime growth to normal processes when this inflation was removed, resulting in a multiplier that is consistent with zero. Another argument is that of Higgs, who remarks that the WW2 economy was essentially a command economy, in which prices were set by government fiat, and thus no aggregate can be computed that can be meaningfully compared to that of 1939, so there is no credible assay showing that recovery occurred prior to 1946.