Debunk­ing Macro­eco­nom­ics

Flattr this!

I have just had the fol­low­ing paper pub­lished in the freely acces­si­ble online jour­nal Eco­nomic Analy­sis and Pol­icy, which is pub­lished by the Queens­land branch of the Eco­nomic Soci­ety of Aus­tralia.The cita­tion is:

Keen, S. (2011). “Debunk­ing Macro­eco­nom­ics.” Eco­nomic Analy­sis & Pol­icy 41(3): 147–167.

I’ve been pub­lished in peer-reviewed jour­nals that are freely acces­si­ble online before of course—such as the Eco­nom­ics E-jour­nal paper “Solv­ing the Para­dox of Mon­e­tary Prof­its”. I think Eco­nomic Analy­sis and Pol­icy does a bet­ter job of blend­ing the old paper-style jour­nal with the new, with a home page that emu­lates the cover of a paper jour­nal while pro­vid­ing hot links to the papers from that page.

The paper can be down­loaded from this link, and you can down­load the whole issue from here. If you enjoy my analy­sis, I rec­om­mend also down­load­ing Mark McGovern’s paper “Beyond the Aus­tralian Debt Dream­time: Recog­nis­ing Imbal­ances”. Here is the abstract to Mark’s paper:

Sadly, all the efforts of a gen­er­a­tion of Aus­tralian men and women have only made them more indebted to the rest of the world. Australia’s exter­nal net wealth is neg­a­tive, soon pass­ing minus $900b on an accel­er­at­ing down­ward tra­jec­tory. This ongo­ing dis­si­pa­tion of national resources is unsus­tain­able. Aus­tralians live in a debt dream­time, one from which the rest of the world has been rudely awak­ened. After years of inad­e­quate poli­cies, the nation has a large exter­nal debt and sig­nif­i­cant gov­ern­ment expo­sures. Ser­vic­ing pres­sures are grow­ing as ris­ing uncer­tain­ties per­me­ate global credit mar­kets. Reserve Bank poli­cies are wors­en­ing Australia’s exter­nal posi­tion and need­lessly dri­ving up inter­nal costs. Major pol­icy rethink­ing is war­ranted. Rel­e­vant issues are still lit­tle con­sid­ered, crowded out of dia­logues by com­fort­ing myths that accom­pany the Aus­tralian Debt Dream­time. Imbal­ances need proper recog­ni­tion with new approaches and strate­gies devel­oped. Auto­matic cor­rec­tions will not occur as his­tory and cur­rent over­seas expe­ri­ences demon­strate. A real awak­en­ing, improved posi­tion­ing and a touch of luck are required if Aus­tralians are to avoid being seri­ously impov­er­ished by world events and their own con­fused Dream­ing.

This paper sum­marises the argu­ments I make at much greater length on macro­eco­nom­ics in the sec­ond edi­tion of Debunk­ing Eco­nom­ics. The one exten­sion to that argu­ment here is a more detailed dis­cus­sion of Hicks’s IS-LM model–and in par­tic­u­lar, Hicks’s admis­sion in 1981 that IS-LM was a pre-Key­ne­sian, neo­clas­si­cal model dat­ing from 1934, which his 1936 “review” of the Gen­eral The­ory falsely passed off as a model of Keynes’s Gen­eral The­ory.


The fail­ure of neo­clas­si­cal mod­els to warn of the eco­nomic cri­sis has led to some rare soul search­ing in a dis­ci­pline not known for such intro­spec­tion. The dom­i­nant reac­tion within the pro­fes­sion has been to admit the fail­ure, but to argue that there is no need for a dras­tic revi­sion of eco­nomic the­ory.

I reject this com­fort­able con­clu­sion, and argue instead that this cri­sis illus­trates the point made before­hand by Robert Solow, that mod­els in which macro­eco­nomic patholo­gies are impos­si­ble are not ade­quate mod­els of cap­i­tal­ism. Hicks’s cri­tique of his own IS-LM model also indi­cates that, though patholo­gies can be imposed on an IS-LM model, it is also inap­pro­pri­ate for macro­eco­nomic analy­sis because of its false impo­si­tion of equi­lib­rium con­di­tions derived from Wal­ras’ Law. I then focus upon what I see as the key weak­ness in the neo­clas­si­cal approach to macro­eco­nom­ics which applies to both DSGE and IS-LM mod­els: the false assump­tion that the money sup­ply is exoge­nous. After out­lin­ing the alter­na­tive endoge­nous money per­spec­tive, I show that Wal­ras’ Law must be gen­er­al­ized for a credit econ­omy to what I call the “Wal­ras-Schum­peter-Min­sky Law”. The empir­i­cal data strongly sup­ports this per­spec­tive, empha­siz­ing the need for a “root and branch” reform of macro­eco­nom­ics.

Defend­ing the inde­fen­si­ble

That mod­ern neo­clas­si­cal macro­eco­nomic mod­els failed to fore­warn of the eco­nomic cri­sis that began in 2007 is undis­puted. What is in dis­pute is the impli­ca­tions this should have for macro­eco­nomic the­ory.

Promi­nent mem­bers of the dis­ci­pline have argued that there should be no con­se­quences. Ben Bernanke (Bernanke 2010), recent Nobel Prize lau­re­ate Thomas Sar­gent (Rol­nick 2010) and the found­ing edi­tor of the AER: Macro Olivier Blan­chard (Blan­chard et al. 2010) have all asserted that neo­clas­si­cal mod­els helped guide pol­icy dur­ing the good times, and should not be aban­doned sim­ply because they did not see the bad times com­ing. Bernanke’s argu­ment is rep­re­sen­ta­tive of this per­spec­tive:

the recent finan­cial cri­sis was more a fail­ure of eco­nomic engi­neer­ing and eco­nomic man­age­ment than of what I have called eco­nomic sci­ence…

Do these fail­ures of stan­dard macro­eco­nomic mod­els mean that they are irrel­e­vant or at least sig­nif­i­cantly flawed? I think the answer is a qual­i­fied no. Eco­nomic mod­els are use­ful only in the con­text for which they are designed. Most of the time, includ­ing dur­ing reces­sions, seri­ous finan­cial insta­bil­ity is not an issue. The stan­dard mod­els were designed for these non-cri­sis peri­ods, and they have proven quite use­ful in that con­text. Notably, they were part of the intel­lec­tual frame­work that helped deliver low infla­tion and macro­eco­nomic sta­bil­ity in most indus­trial coun­tries dur­ing the two decades that began in the mid-1980s.’ (Bernanke 2010, pp. 3, 17; empha­sis added)

I make no apol­ogy for describ­ing this argu­ment as spe­cious, on at least two grounds.

Firstly, this argu­ment would only be tol­er­a­bly accept­able if neo­clas­si­cal eco­nom­ics also had well-devel­oped mod­els that were suit­able for peri­ods of cri­sis, but it does not. Sec­ondly, this blasé accep­tance that there can be bad times sits oddly against the tri­umphal­ism that char­ac­ter­ized neo­clas­si­cal dis­course on macro­eco­nom­ics prior to this cri­sis. This is best exem­pli­fied by Lucas’s Pres­i­den­tial Address to the Amer­i­can Eco­nomic Asso­ci­a­tion in 2003, in which he asserted that neo­clas­si­cal eco­nom­ics had suc­ceeded in elim­i­nat­ing the pos­si­bil­ity of extremely bad times like those we are now expe­ri­enc­ing:

Macro­eco­nom­ics was born as a dis­tinct field in the 1940’s, as a part of the intel­lec­tual response to the Great Depres­sion. The term then referred to the body of knowl­edge and exper­tise that we hoped would pre­vent the recur­rence of that eco­nomic dis­as­ter. My the­sis in this lec­ture is that macro­eco­nom­ics in this orig­i­nal sense has suc­ceeded: Its cen­tral prob­lem of depres­sion pre­ven­tion has been solved, for all prac­ti­cal pur­poses, and has in fact been solved for many decades.’ (Lucas, Jr. 2003, p. 1 ; empha­sis added)

The propo­si­tion that there can be sep­a­rate mod­els for good and bad times also implies that there is no causal link between good and bad times, which would be true if they were sim­ply the prod­uct of exoge­nous shocks to the macro­econ­omy. This is in fact how most neo­clas­si­cal mod­el­ers have reacted: by ret­ro­spec­tively treat­ing the cri­sis as being due, not merely to unprece­dent­edly large exoge­nous shocks, but shocks which var­ied in mag­ni­tude over time—while still remain­ing neg­a­tive rather than pos­i­tive:

the Great Reces­sion began in late 2007 and early 2008 with a series of adverse pref­er­ence and tech­nol­ogy shocks in roughly the same mix and of roughly the same mag­ni­tude as those that hit the United States at the onset of the pre­vi­ous two reces­sions…

The string of adverse pref­er­ence and tech­nol­ogy shocks con­tin­ued, how­ever, through­out 2008 and into 2009. More­over, these shocks grew larger in mag­ni­tude, adding sub­stan­tially not just to the length but also to the sever­ity of the great reces­sion…’ (Ire­land 2011, p. 48, see also McK­ib­bin and Stoeckel 2009)

The fact that these shocks came from the finan­cial sector—which the ordi­nary pub­lic would tend to regard as being part of the econ­omy, and there­fore to be more of an endoge­nous eco­nomic phe­nom­e­non than an exoge­nous event—has been treated as largely irrel­e­vant by lead­ing neo­clas­si­cals:

The cri­sis has shown that large adverse shocks can and do hap­pen. In this cri­sis, they came from the finan­cial sec­tor, but they could come from else­where in the future—the effects of a pan­demic on tourism and trade or the effects of a major ter­ror­ist attack on a large eco­nomic cen­ter.’ (Blan­chard, Dell’Ariccia and Mauro 2010, p. 207)

Given the sever­ity and per­sis­tence of this cri­sis, this is also spe­cious rea­son­ing: if the cri­sis was merely due to a large exoge­nous neg­a­tive shock, surely by now it would be over? Surely too, since the “shocks” emanated from the finance sec­tor, and the stan­dard neo­clas­si­cal doc­trine that per­mits a sep­a­ra­tion of eco­nom­ics from finance has now been empir­i­cally rejected (Fama and French 2004), the treat­ment of dis­tur­bances in the finance sec­tor as “exoge­nous” to the econ­omy should also be rejected?

These reac­tions of neo­clas­si­cal the­o­rists and mod­el­ers are there­fore no more than a plea that the core neo­clas­si­cal vision of the macro­econ­omy as a sta­ble sys­tem sub­ject to exoge­nous shocks should be pre­served, despite an unprece­dented empir­i­cal fail­ure. This propo­si­tion has been put openly by Blan­chard et al., amongst oth­ers:

It is impor­tant to start by stat­ing the obvi­ous, namely, that the baby should not be thrown out with the bath­wa­ter.’ (Blan­chard, Dell’Ariccia and Mauro 2010, p. 204)

How­ever, no lesser Neo­clas­si­cals than John Hicks (Hicks 1981) and Robert Solow (Solow 2003, 2001, 2008) have put the con­trary case that these babies— both today’s DSGE mod­els and the IS-LM model that pre­ceded them—should never have been con­ceived in the first place.

Hicks’s dis­own­ing of IS-LM appears to have dis­ap­peared with­out trace in neo­clas­si­cal lit­er­a­ture. Solow’s voice was at least acknowl­edged by Blan­chard when he pub­lished his excru­ci­at­ingly badly timed sur­vey of mod­ern macro­eco­nom­ics (Blan­chard 2009, Blan­chard 2008), in which he stated that:

after the explo­sion of the field in the 1970s, there has been enor­mous progress and sub­stan­tial con­ver­gence.„ largely because facts have a way of not going away, a largely shared vision both of fluc­tu­a­tions and of method­ol­ogy has emerged… Not every­thing is fine… But none of this is deadly. The state of macro is good.’ (Blan­chard 2009, p. 210; empha­sis added).

In a foot­note, he added “Oth­ers, I know, dis­agree with this opti­mistic assess­ment (for exam­ple, Solow 2008)”, but he did not engage at all with Solow’s cri­tique. As the globe is now in its fifth year of unre­lent­ing eco­nomic tur­moil, it is time Solow was lis­tened to.

Solow’s cri­tique of DSGE

Solow became a critic of neo­clas­si­cal macro­eco­nom­ics because he was sim­ply incred­u­lous that the growth model he devel­oped could be even con­sid­ered as a basis for mod­el­ing the busi­ness cycle:

The pro­to­typ­i­cal real-busi­ness-cycle model … is noth­ing but the neo­clas­si­cal growth model…

The puz­zle I want to discuss—at least it seems to me to be a puz­zle, though part of the puz­zle is why it does not seem to be a puz­zle to many of my younger colleagues—is this. More than forty years ago, I … worked out … neo­clas­si­cal growth the­ory… [I]t was clear from the begin­ning what I thought it did not apply to, namely short-run fluc­tu­a­tions in aggre­gate out­put and employ­ment … the busi­ness cycle…

[N]ow … if you pick up an arti­cle today with the words ‘busi­ness cycle’ in the title, there is a fairly high prob­a­bil­ity that its basic the­o­ret­i­cal ori­en­ta­tion will be what is called ‘real busi­ness cycle the­ory’ and the under­ly­ing model will be … a slightly dressed up ver­sion of the neo­class­si­cal growth model. The ques­tion I want to cir­cle around is: how did that hap­pen?’ (Solow 2001, pp. 23, 19)

Solow iden­ti­fied the search for micro­foun­da­tions for macro­eco­nom­ics as the foun­tain­head of DSGE mod­el­ing, and though he sur­mised that “The orig­i­nal impulse to look for bet­ter or more explicit micro foun­da­tions was prob­a­bly rea­son­able” (Solow 2003, p. 1), he dis­missed the defense that micro­eco­nom­ics jus­ti­fied DSGE macro­eco­nomic mod­els as “a delu­sion”:

Sup­pose you wanted to defend the use of the Ram­sey model as the basis for a descrip­tive macro­eco­nom­ics. What could you say?… (I take it for granted that “real­ism” is not an eli­gi­ble defense.)

You could claim that it is not pos­si­ble to do bet­ter at this level of abstrac­tion; that there is no other tractable way to meet the claims of eco­nomic the­ory. I think this claim is a delu­sion.

We know from the Son­nen­schein-Man­tel-Debreu the­o­rems that the only uni­ver­sal empir­i­cal aggrega­tive impli­ca­tions of gen­eral equi­lib­rium the­ory are that excess demand func­tions should be con­tin­u­ous and homo­ge­neous of degree zero in prices, and should sat­isfy Wal­ras’ Law. Any­one is free to impose fur­ther restric­tions on a macro model, but they have to be jus­ti­fied for their own sweet sake, not as being required by the prin­ci­ples of eco­nomic the­ory.’ (Solow 2008, pp. 244)

Solow rejected both the “Salt­wa­ter” and “Fresh­wa­ter” approaches to macro­eco­nom­ics. The base model itself, the “Fresh­wa­ter” real busi­ness cycle model, was unsuit­able for macro­eco­nom­ics because it ruled out the very behav­ior that macro­eco­nom­ics is sup­posed to explain:

The pre­ferred model has a sin­gle rep­re­sen­ta­tive con­sumer opti­miz­ing over infi­nite time with per­fect fore­sight or ratio­nal expec­ta­tions, in an envi­ron­ment that real­izes the result­ing plans more or less flaw­lessly through per­fectly com­pet­i­tive for­ward-look­ing mar­kets for goods and labor, and per­fectly flex­i­ble prices and wages.

How could any­one expect a sen­si­ble short-to-medium-run macro­eco­nom­ics to come out of that set-up?…

I start from the pre­sump­tion that we want macro­eco­nom­ics to account for the occa­sional aggrega­tive patholo­gies that beset mod­ern cap­i­tal­ist economies, like reces­sions, inter­vals of stag­na­tion, infla­tion, “stagfla­tion,” not to men­tion neg­a­tive patholo­gies like unusu­ally good times. A model that rules out patholo­gies by def­i­n­i­tion is unlikely to help.’ (Solow 2003, p. 1; empha­sis added).

He dis­missed the “New Key­ne­sian” vari­ant that came to dom­i­nate the pro­fes­sion as no more than win­dow-dress­ing to improve the appar­ent fit of a bad model to the data:

The sim­pler sort of RBC model that I have been using for expos­i­tory pur­poses has had lit­tle or no empir­i­cal suc­cess, even with a very unde­mand­ing notion of ‘empir­i­cal suc­cess’. As a result, some of the freer spir­its in the RBC school have begun to loosen up the basic frame­work by allow­ing for ‘imper­fec­tions’ in the labor mar­ket, and even in the cap­i­tal mar­ket…

The model then sounds bet­ter and fits the data bet­ter. This is not sur­pris­ing: these imper­fec­tions were cho­sen by intel­li­gent econ­o­mists to make the mod­els work bet­ter...’ (Solow 2001, p. 26; empha­sis added)

One need not won­der how Solow would react to neo­clas­si­cal macro­econ­o­mists, after the cri­sis, adding not merely imper­fec­tions but adjustable exoge­nous shocks to improve the model’s fit to data that “imper­fec­tions” alone can­not explain, since he gave his opin­ion before­hand:

It is always pos­si­ble to claim that those “patholo­gies” are delu­sions, and the econ­omy is merely adjust­ing opti­mally to some exoge­nous shock. But why should rea­son­able peo­ple accept this?’ (Solow, 2003, p. 1; empha­sis added)

Though Solow saw the des­ti­na­tion that neo­clas­si­cal macro­eco­nom­ics had reached as an impasse, he pro­vided no alter­nate route for­ward. Some hark for a rever­sal of direc­tion back to IS-LM mod­els (Man­fred Gärt­ner and Flo­rian Jung, 2010), in which patholo­gies at least appear fea­si­ble. How­ever that route was blocked three decades ago by one John Hicks.

Hicks’s Cri­tique of IS-LM

In 1981, John Hicks admit­ted that, though regarded as a model of Keynes, IS-LM is neo­clas­si­cal model that pre­dates Keynes’s Gen­eral The­ory (J. M. Keynes, 1936). Hicks noted that though he first spelled out the IS-LM model in detail in “Mr. Keynes and the Clas­sics” (J. R. Hicks, 1937, his review of The Gen­eral The­ory), the model was first devel­oped in a less well-known paper, “Wages and Inter­est: The Dynamic Prob­lem” (J. R. Hicks, 1935). Hicks was adamant that this paper, and not “Mr. Keynes and the Classics”—let alone The Gen­eral The­ory itself—was the real foun­da­tion of the IS-LM model:

The other, much less well known, is even more rel­e­vant. “Wages and Inter­est: the Dynamic Prob­lem” was a first sketch of what was to become the “dynamic” model of Value and Cap­i­tal (1939). It is impor­tant here, because it shows (I think quite con­clu­sively) that that model was already in my mind before I wrote even the first of my papers on Keynes.’ (Hicks 1981, p. 140; empha­sis added)

Of course, that IS-LM is in fact a Neo­clas­si­cal model and not a Key­ne­sian one is no rea­son per se to dis­miss it: it could still be an ade­quate model of the macro econ­omy. But Hicks argued that on its own mer­its, it failed.

The model in “Wages and Inter­est” had an ultra-short-run of a week in a “bread econ­omy” in which prices were decided on a Mon­day and then applied for the remain­der of the week. But IS-LM, which Hicks described as “a trans­la­tion of Keynes’ non­flex­price model into my terms”, was a ““short-period,” … we shall not go far wrong if we think of it as a year” (John Hicks, 1981, p. 141). In Hicks’s model, events dur­ing the week were not allowed to affect anything—a device which Hicks described as “a very arti­fi­cial device, not (I would think now) much to be rec­om­mended”, but which let him treat expec­ta­tions as con­stant. There­fore, equi­lib­rium applied:

That is to say (it is equiv­a­lent to say­ing), we may fairly reckon that these mar­kets, with respect to these data, are in equi­lib­rium.’ (Hicks 1981, p. 146)

How­ever, this arti­fice could not be extended to a year, and used to derive an LM curve since:

for the pur­pose of gen­er­at­ing an LM curve, which is to rep­re­sent liq­uid­ity pref­er­ence, it will not do with­out amend­ment. For there is no sense in liq­uid­ity, unless expec­ta­tions are uncer­tain.’ (Hicks 1981, p. 152; empha­sis added)

This in turn meant that mar­kets had to be in disequilibrium—but IS-LM itself was derived from equi­lib­rium analy­sis, as Hicks also explained:

the idea of the IS-LM dia­gram came to me as a result of the work I had been doing on three-way exchange, con­ceived in a Wal­rasian man­ner. I had already found a way of rep­re­sent­ing three-way exchange on a two-dimen­sional dia­gram (to appear in due course in chap­ter 5 of Value and Cap­i­tal). As it appears there, it is a piece of sta­t­ics; but it was essen­tial to my approach (as already appears in “Wages and Inter­est: the Dynamic Prob­lem”) that sta­tic analy­sis of this sort could be car­ried over to “dynam­ics” by rede­f­i­n­i­tion of terms. So it was nat­ural for me to think that a sim­i­lar device could be used for the Keynes the­ory.’ (Hicks 1981, p. 141–142)

But in fact, extend­ing an equi­lib­rium model with a time­frame of a week to a time­frame of a year in which liq­uid­ity pref­er­ence played a key role led to a hope­less mud­dle which Hicks, with hind­sight, could now appre­ci­ate. His final judg­ment on his own model was scathing:

I accord­ingly con­clude that the only way in which IS-LM analy­sis use­fully survives—as any­thing more than a class­room gad­get, to be super­seded, later on, by some­thing better—is in appli­ca­tion to a par­tic­u­lar kind of causal analy­sis, where the use of equi­lib­rium meth­ods, even a dras­tic use of equi­lib­rium meth­ods, is not inap­pro­pri­ate…

When one turns to ques­tions of pol­icy … the use of equi­lib­rium meth­ods is still more sus­pect. … There can be no change of pol­icy if every­thing is to go on as expected—if the econ­omy is to remain in what (how­ever approx­i­mately) may be regarded as its exist­ing equi­lib­rium. It may be hoped that, after the change in pol­icy, the econ­omy will some­how, at some time in the future, set­tle into what may be regarded, in the same sense, as a new equi­lib­rium; but there must nec­es­sar­ily be a stage before that equi­lib­rium is reached. There must always be a prob­lem of tra­verse. For the study of a tra­verse, one has to have recourse to sequen­tial meth­ods of one kind or another.’ (Hicks 1981, p. 152–153)

Hicks’s fun­da­men­tal con­clu­sion there­fore is that macro­eco­nomic analy­sis must assume dis­e­qui­lib­rium rather than equilibrium—and this does even more dam­age to IS-LM than Hicks him­self appre­ci­ated. To reduce macro­eco­nom­ics to the inter­play of sim­ply the goods and money mar­ket, Hicks had used the Wal­rasian assump­tion that, if n-1 mar­kets are in equi­lib­rium, then the nth mar­ket must also be in equi­lib­rium, so that he could ignore the mar­ket for loan­able funds:

One did not have to bother about the mar­ket for “loan­able funds,” since it appeared, on the Wal­ras anal­ogy, that if these two “mar­kets” were in equi­lib­rium, the third must be also. So I con­cluded that the inter­sec­tion of IS and LM deter­mined the equi­lib­rium of the sys­tem as a whole.’ (Hicks 1981, p. 142)

But this Wal­rasian logic cuts both ways: if dis­e­qui­lib­rium applies in one mar­ket, then at least one other—and prob­a­bly all others—must also be in dis­e­qui­lib­rium. There­fore, as soon as dis­e­qui­l­brium is acknowl­edged, mar­kets whose very exis­tence has been ignored in equi­lib­rium analysis—such as, obvi­ously in IS-LM, the labor market—can no longer be ignored. Their dis­e­qui­l­brium dynam­ics must now be con­sid­ered, and the IS-LM model can give no guid­ance as to how they will behave.

An alter­na­tive macro­eco­nom­ics

Solow’s key obser­va­tion is that macro­eco­nom­ics should seek to “account for the occa­sional aggrega­tive patholo­gies that beset mod­ern cap­i­tal­ist economies”. Hicks’s key con­tri­bu­tion was that macro­eco­nom­ics must be a study of dis­e­qui­l­brium dynamics—a point made decades ear­lier still by Irv­ing Fisher when he penned “The Debt-Defla­tion The­ory of Great Depres­sions”:

We may ten­ta­tively assume that, ordi­nar­ily and within wide lim­its, all, or almost all, eco­nomic vari­ables tend, in a gen­eral way, toward a sta­ble equi­lib­rium… But … New dis­tur­bances are, humanly speak­ing, sure to occur, so that, in actual fact, any vari­able is almost always above or below the ideal equi­lib­rium…

The­o­ret­i­cally there may be—in fact, at most times there must be—over-or under-pro­duc­tion, over- or under-con­sump­tion, over- or under-spend­ing, over- or under-sav­ing, over- or under-invest­ment, and over or under every­thing else. It is as absurd to assume that, for any long period of time, the vari­ables in the eco­nomic orga­ni­za­tion, or any part of them, will “stay put,” in per­fect equi­lib­rium, as to assume that the Atlantic Ocean can ever be with­out a wave.’ (Fisher 1933, p. 339; empha­sis added)

These prin­ci­ples point in the direc­tion first indi­cated by Hyman Min­sky, that since cap­i­tal­ist economies have expe­ri­enced Depres­sions in the past, to ade­quately model cap­i­tal­ism:

it is nec­es­sary to have an eco­nomic the­ory which makes great depres­sions one of the pos­si­ble states in which our type of cap­i­tal­ist econ­omy can find itself.’ (Min­sky 1982 , p. 5)

Minsky’s alter­na­tive was the “Finan­cial Insta­bil­ity Hypoth­e­sis”, and it was based on an explicit rejec­tion of the neo­clas­si­cal par­a­digm:

The abstract model of the neo­clas­si­cal syn­the­sis can­not gen­er­ate insta­bil­ity. When the neo­clas­si­cal syn­the­sis is con­structed, cap­i­tal assets, financ­ing arrange­ments that cen­ter around banks and money cre­ation, con­straints imposed by lia­bil­i­ties, and the prob­lems asso­ci­ated with knowl­edge about uncer­tain futures are all assumed away. For econ­o­mists and pol­icy-mak­ers to do bet­ter we have to aban­don the neo­clas­si­cal syn­the­sis.’ (Min­sky 1982 , p. 5)

I have devel­oped math­e­mat­i­cal mod­els of this hypoth­e­sis (Steve Keen, 1995, 2008, 2011b, 2000) which are based on a con­sid­ered rejec­tion of vir­tu­ally every pre­cept of neo­clas­si­cal the­ory. Explain­ing all these method­olog­i­cal deci­sions requires a book rather than a jour­nal paper (Steve Keen, 2011a), so here I will con­cen­trate on what expe­ri­ence has con­vinced me is the key rea­son why neo­clas­si­cal econ­o­mists failed to fore­see the cri­sis, and why to this day they can­not under­stand why it remains so intractable. This is their vision of how money is cre­ated.

Endoge­nous Money, Eco­nomic Growth and Dis­e­qui­lib­rium

Non-econ­o­mists might expect that eco­nomic mod­els of how money is cre­ated would be based on empir­i­cal research. There is such a model, but it is the province of the non-neo­clas­si­cal school of thought known as Post Key­ne­sian eco­nom­ics. Neo­clas­si­cal econ­o­mists have instead per­sisted with the “frac­tional reserve banking/money mul­ti­plier” model, which argues that banks need excess reserves before they can lend, that these are cre­ated ini­tially by an expan­sion of gov­ern­ment-cre­ated “fiat” money, and that a sequence of bank deposits by recip­i­ents of fiat money, and loans of all but a frac­tion of this deposit by banks, cre­ates credit money that is a mul­ti­ple of the ini­tial injec­tion of fiat money.

This is in turn mar­ried to a vision of banks as mere inter­me­di­aries, so that they can be for­mally ignored in macro­eco­nomic mod­el­ing. Finally, the level of pri­vate debt is also ignored in neo­clas­si­cal macro­eco­nom­ics, since a loan is regarded as sim­ply a trans­fer of spend­ing power from a saver to a bor­rower. With one person’s spend­ing power going down and another’s going up by the same amount, only the dis­tri­b­u­tion of debt could matter—not its aggre­gate level. As Bernanke explained, this is why Fisher’s “Debt Defla­tion” expla­na­tion of the Great Depres­sion was ignored by neo­clas­si­cal econ­o­mists:

Fisher’s idea was less influ­en­tial in aca­d­e­mic cir­cles, though, because of the coun­ter­ar­gu­ment that debt-defla­tion rep­re­sented no more than a redis­tri­b­u­tion from one group (debtors) to another (cred­i­tors). Absent implau­si­bly large dif­fer­ences in mar­ginal spend­ing propen­si­ties among the groups, it was sug­gested, pure redis­tri­b­u­tions should have no sig­nif­i­cant macro-eco­nomic effects…’ (Bernanke 2000, p. 24)

Over forty years ago, the then Senior Vice-Pres­i­dent of the New York Fed­eral Reserve, Alan Holmes, pointed out that this per­spec­tive in which the aggre­gate level of debt does not mat­ter, and banks are mere inter­me­di­aries between savers and lenders, was erro­neous. He argued that the view “that the bank­ing sys­tem only expands loans after the [Fed­eral Reserve] Sys­tem (or mar­ket fac­tors) have put reserves in the bank­ing sys­tem” was based on “a naive assump­tion”. Instead, he argued,

In the real world, banks extend credit, cre­at­ing deposits in the process, and look for the reserves later. The ques­tion then becomes one of whether and how the Fed­eral Reserve will accom­mo­date the demand for reserves. In the very short run, the Fed­eral Reserve has lit­tle or no choice about accom­mo­dat­ing that demand; over time, its influ­ence can obvi­ously be felt.’ (Holmes 1969, p. 73; empha­sis added)

Holmes was thus argu­ing that banks do not need deposits in order to lend—that in fact the act of lend­ing simul­ta­ne­ously cre­ates a match­ing deposit—and that banks can there­fore endoge­nously cre­ate new spend­ing power (new deposits) by issu­ing a loan, with­out thereby reduc­ing the spend­ing power of savers.

This expe­ri­ence-based judg­ment was sub­se­quently con­firmed by empir­i­cal research by the Post Key­ne­sian econ­o­mist Basil Moore (Basil J. Moore, 1979, 1988, 2001, 1983), and even by the founders of Real Busi­ness Cycle the­ory (Finn E. Kyd­land and Edward C. Prescott, 1990). This led to the devel­op­ment of the model of “endoge­nous money”, in which a loan is regarded not a trans­fer of spend­ing power from a saver to a lender, but a cre­ation of spend­ing power for the bor­rower by the bank ab ini­tio. This mod­ern Post Key­ne­sian the­ory in fact redis­cov­ered an argu­ment first put by Schum­peter, that banks cre­ate money sim­ply by an account­ing oper­a­tion: a loan extended to a bor­rower cre­ates both debt and spend­ing power “out of noth­ing”:

It is always a ques­tion, not of trans­form­ing pur­chas­ing power which already exists in someone’s pos­ses­sion, but of the cre­ation of new pur­chas­ing power out of noth­ing…’ (Schum­peter 1934, p. 73)

This means that the increase in debt actu­ally adds to aggre­gate demand, so that total spend­ing is the sum of both incomes gen­er­ated in “the cir­cu­lar flow” —which pri­mar­ily finances consumption—plus the growth in debt—which pri­mar­ily finances invest­ment. Since this con­cept is so for­eign to neo­clas­si­cal econ­o­mists, it is worth cit­ing Schum­peter at length on it here:

the entrepreneur—in prin­ci­ple and as a rule—does need credit, in the sense of a tem­po­rary trans­fer to him of pur­chas­ing power, in order to pro­duce at all, to be able to carry out his new com­bi­na­tions, to become an entre­pre­neur. And this pur­chas­ing power does not flow towards him auto­mat­i­cally, as to the pro­ducer in the cir­cu­lar flow, by the sale of what he pro­duced in pre­ced­ing peri­ods. If he does not hap­pen to pos­sess it … he must bor­row it… He can only become an entre­pre­neur by pre­vi­ously becom­ing a debtor… his becom­ing a debtor arises from the neces­sity of the case and is not some­thing abnor­mal, an acci­den­tal event to be explained by par­tic­u­lar cir­cum­stances. What he first wants is credit. Before he requires any goods what­ever, he requires pur­chas­ing power. He is the typ­i­cal debtor in cap­i­tal­ist soci­ety.’ (see also Biggs and Mayer 2010, Biggs et al. 2010, Schum­peter 1934, p. 102)

Schum­peter does not deny that some invest­ment is financed by a trans­fer of exist­ing funds from savers—hence result­ing in a fall in spend­ing by saver-con­sumers and an off­set­ting increase in spend­ing by bor­rower-investors, with no over­all macro­eco­nomic impli­ca­tions. But he insists that the pri­mary source of investor spend­ing comes from the endoge­nous expan­sion of the money sup­ply, which gives spend­ing power to entre­pre­neurs with­out sac­ri­fic­ing the exist­ing spend­ing power of savers, so that ris­ing debt does have macro­eco­nomic effects:

Even though the con­ven­tional answer to our ques­tion is not obvi­ously absurd, yet there is another method of obtain­ing money for this pur­pose, which … does not pre­sup­pose the exis­tence of accu­mu­lated results of pre­vi­ous devel­op­ment, and hence may be con­sid­ered as the only one which is avail­able in strict logic. This method of obtain­ing money is the cre­ation of pur­chas­ing power by banks…’ (Schum­peter 1934, p. 73)

This leads Schum­peter to what he hap­pily describes as a “heresy… that processes in terms of means of pay­ment are not merely reflexes of processes in terms of goods” (Joseph Alois Schum­peter, 1934, p. 95). Instead, in a grow­ing cap­i­tal­ist econ­omy, aggre­gate demand is the sum of demand from the sale of goods and ser­vices plus demand from the growth in credit-money, where the lat­ter is the pri­mary source of invest­ment. Fama and French con­firmed this hypoth­e­sized rela­tion­ship between change in debt and invest­ment in an empir­i­cal study:

These cor­re­la­tions con­firm the impres­sion that debt plays a key role in accom­mo­dat­ing year-by-year vari­a­tion in invest­ment.’ (Fama and French 1999, p. 1954)

This com­bi­na­tion of the endoge­nous expan­sion of spend­ing power by bank lend­ing, plus the use of that by entre­pre­neurs to finance invest­ment, means that the change in the level of pri­vate debt does have macro­eco­nomic significance—and it plays a cen­tral role in Schumpeter’s the­ory of the busi­ness cycle. How­ever it is not the end of the mat­ter, because entre­pre­neurs are not the only ones who bor­row money: so do key actors in Minsky’s expla­na­tion for Great Depres­sions, “Ponzi Financiers”.

These bor­row­ers do not pri­mar­ily invest with bor­rowed money, but buy exist­ing assets, and hope to profit by sell­ing those assets on a ris­ing mar­ket. Unlike Schumpeter’s entre­pre­neurs, whose debts today can be repaid from prof­its tomor­row, Ponzi Financiers always have debt ser­vic­ing costs than exceed the cash flows from the assets they pur­chase with bor­rowed money. They there­fore must expand their debts or sell assets to con­tinue func­tion­ing:

A Ponzi finance unit is a spec­u­la­tive financ­ing unit for which the income com­po­nent of the near term cash flows falls short of the near term inter­est pay­ments on debt so that for some time in the future the out­stand­ing debt will grow due to inter­est on exist­ing debt… Ponzi units can ful­fill their pay­ment com­mit­ments on debts only by bor­row­ing (or dis­pos­ing of assets)… a Ponzi unit must increase its out­stand­ing debts.’ (Min­sky 1982, p. 24)

There­fore in a credit-based econ­omy, there are three sources of aggre­gate demand, and three ways in which this demand is expended:

  1. Demand from income earned by sell­ing goods and ser­vices, which pri­mar­ily finances con­sump­tion;
  2. Demand from ris­ing entre­pre­neur­ial debt, which pri­mar­ily finances invest­ment; and
  3. Demand from ris­ing Ponzi debt, which pri­mar­ily finances the pur­chase of exist­ing assets.

Schumpeter’s and Minsky’s per­spec­tives thus enable us to inte­grate credit, asset mar­kets and dis­e­qui­lib­rium analy­sis into an alter­na­tive macro­eco­nom­ics. In a credit econ­omy, aggre­gate demand is the sum of income plus the change in debt, and this demand is expended on both goods and ser­vices and pur­chases of exist­ing assets. Debt there­fore has both pos­i­tive and neg­a­tive con­no­ta­tions for the econ­omy: it finances the expan­sion of eco­nomic activ­ity via inno­va­tion and invest­ment, but it can also cause asset bub­bles and even­tu­ally, an eco­nomic cri­sis if too much of this debt is directed to Ponzi Finance. This is pre­cisely what occurred in the last two decades.

Com­par­ing Endoge­nous Money to “the Money Mul­ti­plier”

As is well known, Bernanke blamed the Fed for caus­ing the Great Depres­sion:

there is now over­whelm­ing evi­dence that the main fac­tor depress­ing aggre­gate demand was a world­wide con­trac­tion in world money sup­plies. This mon­e­tary col­lapse was itself the result of a poorly man­aged and tech­ni­cally flawed inter­na­tional mon­e­tary sys­tem (the gold stan­dard, as recon­sti­tuted after World War I).’ (Bernanke 2000, p. ix)

The money mul­ti­plier the­ory of money cre­ation played a large role in his argu­ment that the Great Depres­sion was trig­gered by the Fed­eral Reserve’s reduc­tion of the US base money sup­ply between June 1928 and June 1931:

the United States is the only coun­try in which the dis­cre­tionary com­po­nent of pol­icy was arguably sig­nif­i­cantly desta­bi­liz­ing… the ratio of mon­e­tary base to inter­na­tional reserves … fell con­sis­tently in the United States from … 1928:II … through the sec­ond quar­ter of 1931. As a result, U.S. nom­i­nal money growth was pre­cisely zero between 1928:IV and 1929:IV, despite both gold inflows and an increase in the money mul­ti­plier.

The year 1930 was even worse in this respect: between 1929:IV and 1930:IV, nom­i­nal money in the United States fell by almost 6 [per­cent]… The prox­i­mate cause of this decline in M1 was con­tin­ued con­trac­tion in the ratio of base to reserves, which rein­forced rather than off­set declines in the money mul­ti­plier. This tight­en­ing … locates much of the blame for the early (pre-1931) slow­down in world mon­e­tary aggre­gates with the Fed­eral Reserve.’ (Ben S. Bernanke, 2000, p. 153)

Bernanke’s sta­tis­ti­cal evi­dence focused on the rela­tion­ship between change in M1 and the level of unem­ploy­ment. Using a longer and more detailed series for M1 than Bernanke used (Fried­man and Schwartz 1963, Table A1), there is indeed a robust neg­a­tive cor­re­la­tion between change in M1 and unem­ploy­ment (see Fig­ure 1; unem­ploy­ment is inverted on the right hand axis to show the neg­a­tive cor­re­la­tion more clearly). The R2 is –0.33 for the whole period, –0.31 for 1920–1930 and a strong –0.67 for 1930–1940.

Fig­ure 1: Change in M1 and Unem­ploy­ment, 1920–1940

How­ever M1 includes money cre­ated by the banks, and blam­ing the Fed­eral Reserve for its col­lapse in the 1930s takes the money mul­ti­plier for granted as a valid expla­na­tion of money creation—which the endoge­nous money approach dis­putes. Only M0 or Base Money is strictly under the Fed’s con­trol, and (using the St Louis FRED AMBSL data), the rela­tion­ship between unem­ploy­ment and that part of the money sup­ply over which the Fed has undis­puted con­trol is very dif­fer­ent (see Fig­ure 2). The cor­re­la­tion for the whole time period now has the wrong sign (+0.44); more­over there is a break in the rela­tion­ship. In the period 1920–1930, the cor­re­la­tion is small with the cor­rect sign (-0.22); in the period 1930–1940, it is small with the wrong sign (+0.28).

Fig­ure 2: Change in Base Money and Unem­ploy­ment, 1920–1940

There is also a break in the rela­tion­ship between changes in M0 and changes in M1. In the 1920s, the cor­re­la­tion was strong and pos­i­tive (+0.89)—in line with the “money mul­ti­plier” argu­ment. But in the 1930s, while still pos­i­tive it was much weaker (+0.34). From the data too, it is obvi­ous that the 1930s Fed was try­ing to boost the money sup­ply from 1931 till 1937. It then reduced its stim­u­lus in 1937 when it falsely believed that the cri­sis was over as unem­ploy­ment began to fall, only to see it start to rise once again from 11 back to 20 per­cent. Shocked, it rapidly reversed direc­tion, increas­ing base money by more than 20 per­cent per annum from mid-1938 on. Arguably the direc­tion of cau­sa­tion was not that changes in M0 and M1 drive unem­ploy­ment, but that changes in unem­ploy­ment drive Fed­eral Reserve pol­icy and hence changes in M0—but with lit­tle impact on the over­all growth of the money sup­ply and hence lit­tle impact upon eco­nomic activ­ity.

Clearly the “The Fed Did It” argu­ment is on shaky grounds with respect to the Great Depres­sion, and as Bernanke is now find­ing out the hard way, it is an even more sus­pect argu­ment today. Firstly the rela­tion­ship of growth in M3 to unem­ploy­ment is even stronger than the M1 to unem­ploy­ment rela­tion from 1920–1940: the R2 is –0.7 for the whole period.

Fig­ure 3: M3 Change and Unem­ploy­ment are strongly neg­a­tively cor­re­lated (R^2 = –0.72)

How­ever the M0 to unem­ploy­ment cor­re­la­tion has the wrong sign, and is even stronger than the wrong cor­re­la­tion for the 1930s at +0.51 (see Fig­ure 4). The biggest boost to Base Money in recorded his­tory did lit­tle to avert the col­lapse into the Great Reces­sion, and repeated boosts are doing very lit­tle to end it.

Fig­ure 4: The Cor­re­la­tion of M0 Change and Unem­ploy­ment has the “Wrong” Sign

The “Money Mul­ti­plier” also has even less effect on the money sup­ply itself now than back in the Great Depres­sion: the cor­re­la­tion between changes in M0 and changes in M3 is effec­tively zero for the whole period since 1990 (-0.07), strongly of the wrong sign for the pre-cri­sis period from 1990–2008 (-0.55), and only barely pos­i­tive for the post-cri­sis period (+0.14).

Fig­ure 5: Change in M0 has no Cor­re­la­tion with Change in M3

In con­trast, the endoge­nous money analy­sis of both the Great Depres­sion and the “Great Con­trac­tion” ( is sub­stan­tially more robust. To test the endoge­nous money expla­na­tion, I put the propo­si­tion that aggre­gate demand is the sum of income plus the change in debt, and that this is expended on both goods and ser­vices and pur­chases of exist­ing assets, in an equa­tion where the left hand side rep­re­sents mon­e­tary flows and the right hand side rep­re­sents the mon­e­tary value of phys­i­cal sup­plies and turnover of finan­cial claims on phys­i­cal assets:


NAT stands for “Net Asset Turnover”, which can be fac­tored into the price index for assets PA, times their quan­tity QA, times the turnover TA expressed as a frac­tion of the num­ber of assets 


The endoge­nous money hypoth­e­sis thus asserts that there is a rela­tion­ship between the change in debt and the level of both eco­nomic activ­ity (GDP) and activ­ity on asset mar­kets. Since the strength of this rela­tion­ship vis-a-vis the impact of income on eco­nomic activ­ity and asset mar­kets depends on the size of the change in debt rela­tion to the level of income, in the fol­low­ing cor­re­la­tions I con­sider change in debt rel­a­tive to GDP rather than change in debt per se.

The cor­re­la­tion between change in debt and unem­ploy­ment is far stronger than the cor­re­la­tion between change in M1 and unem­ploy­ment dur­ing both peri­ods:

Fig­ure 6: Change in Debt Dri­ves Eco­nomic Per­for­mance

Fig­ure 7: Change in Debt Dri­ves Eco­nomic Per­for­mance

Sim­i­larly, there is a rela­tion­ship between the accel­er­a­tion of debt and both the rate of change of GDP (Biggs and Mayer 2010, Biggs, Mayer and Pick 2010) and the change in asset prices:



This rela­tion can be seen as a gen­er­al­iza­tion of Bernanke’s “Finan­cial Accel­er­a­tor” (Bernanke et al. 1996) to both asset mar­kets as well as goods mar­kets, and with­out the false Neo­clas­si­cal restric­tion that only the dis­tri­b­u­tion of debt, and not its aggre­gate rate of change, has macro­eco­nomic sig­nif­i­cance.

The data strongly sup­port the endoge­nous money propo­si­tion that the accel­er­a­tion of debt has macro­eco­nomic sig­nif­i­cance. The cor­re­la­tion between credit accel­er­a­tion and change in unem­ploy­ment in the 1920–40 period is sig­nif­i­cant and has the cor­rect sign.

Fig­ure 8: Debt Accel­er­a­tion and Unem­ploy­ment Change, 1920–1940

The cor­re­la­tion for the period 1990-Now is stronger still (-0.75) and in fact is larger than the cor­re­la­tion of change in M3 with the level of unemployment—let alone its rate of change.

Fig­ure 9: Debt Accel­er­a­tion and Unem­ploy­ment Change, 1990–2012

The cor­re­la­tion of the accel­er­a­tion of debt with change in asset prices is also sub­stan­tial. In strong con­trast to the now dis­cred­ited Effi­cient Mar­kets Hypoth­e­sis claim that debt plays no role in deter­min­ing the value of even a sin­gle cor­po­ra­tion, the accel­er­a­tion of debt has a sub­stan­tial influ­ence upon the change in val­u­a­tion of the entire asset mar­ket.

Fig­ure 10: Credit Accel­er­a­tion & the DJIA, 1922–1940

Fig­ure 11: Credit Accel­er­a­tion and the DJIA, 1990-Now

This is espe­cially marked in the case of hous­ing, when the accel­er­a­tion of mort­gage debt can be iso­lated. The cor­re­la­tion of the accel­er­a­tion of mort­gage debt with the change in real house prices is 0.8.

Fig­ure 12: Mort­gage Accel­er­a­tion and Change in Real House Prices

Since debt can­not accel­er­ate indef­i­nitely, this is also why asset bub­bles, ulti­mately, have to crash—and there­fore also a rea­son why macro­eco­nomic pol­icy should attempt to pre­vent them in the first place.

Con­clu­sion: towards a new macro­eco­nom­ics

The analy­sis here is still only pre­lim­i­nary, but the data gives far stronger empir­i­cal sup­port to the endoge­nous money, credit-dri­ven per­spec­tive than to the neo­clas­si­cal. For both the­o­ret­i­cal and empir­i­cal rea­sons, Neo­clas­si­cal macro­eco­nom­ics does indeed deserve to be “thrown out with the bath­wa­ter”, and a new dynamic, dis­e­qui­lib­rium macro­eco­nom­ics con­structed in its stead.

Fail­ure to hon­estly con­front the the­o­ret­i­cal weak­nesses and empir­i­cal fail­ures of the dom­i­nant school of eco­nom­ics will not pre­serve it. Indeed, con­tin­u­ing to ignore the gulf between the pre­dic­tions of neo­clas­si­cal eco­nomic mod­els and the state of the econ­omy will sim­ply accel­er­ate the steep decline in the pub­lic respect accorded to econ­o­mists since this cri­sis began. Ulti­mately, com­pla­cency and denial will lead eco­nom­ics back to the posi­tion of dis­dain that Keynes described so well dur­ing the last Depres­sion:

But although the doc­trine itself has remained unques­tioned by ortho­dox econ­o­mists up to a late date, its sig­nal fail­ure for pur­poses of sci­en­tific pre­dic­tion has greatly impaired, in the course of time, the pres­tige of its prac­ti­tion­ers. For pro­fes­sional econ­o­mists, after Malthus, were appar­ently unmoved by the lack of cor­re­spon­dence between the results of their the­ory and the facts of obser­va­tion; a dis­crep­ancy which the ordi­nary man has not failed to observe, with the result of his grow­ing unwill­ing­ness to accord to econ­o­mists that mea­sure of respect which he gives to other groups of sci­en­tists whose the­o­ret­i­cal results are con­firmed by obser­va­tion when they are applied to the facts. (J. M. Keynes, 1936, p. 33)


Bernanke, B. (2010). On the Impli­ca­tions of the Finan­cial Cri­sis for Eco­nom­ics. Con­fer­ence Co-spon­sored by the Cen­ter for Eco­nomic Pol­icy Stud­ies and the Bend­heim Cen­ter for Finance, Prince­ton Uni­ver­sity. Prince­ton, NJ: US Fed­eral Reserve.

Bernanke, B.S., M. Gertler and S. Gilchrist (1996). The Finan­cial Accel­er­a­tor and the Flight to Qual­ity, Review of Eco­nom­ics and Sta­tis­tics. 78: 1–15.

Bernanke, B.S. (2000). Essays on the Great Depres­sion. Prince­ton, NJ: Prince­ton Uni­ver­sity Press.

Biggs, M. and T. Mayer (2010). The Out­put Gap Conun­drum, Intereconomics/Review of Euro­pean Eco­nomic Pol­icy. 45: 11–16.

Biggs, M., T. Mayer, and A. Pick (2010). Credit and Eco­nomic Recov­ery: Demys­ti­fy­ing Phoenix Mir­a­cles. SSRN eLi­brary.

Blan­chard, O. (2008). The State of Macro, National Bureau of Eco­nomic Research Work­ing Paper Series, No. 14259.

Blan­chard, O. (2009). The State of Macro, Annual Review of Eco­nom­ics. 1, 209–28.

Blan­chard, O., G. Dell’Ariccia, and P. Mauro (2010). Rethink­ing Macro­eco­nomic Pol­icy, Jour­nal of Money, Credit, and Bank­ing. 42: 199–215.

Fama, E.F. and K.R. French (1999). The Cor­po­rate Cost of Cap­i­tal and the Return on Cor­po­rate Invest­ment, Jour­nal of Finance. 54: 1939–67.

Fama, E.F. and K.R. French (2004). The Cap­i­tal Asset Pric­ing Model: The­ory and Evi­dence, The Jour­nal of Eco­nomic Per­spec­tives. 18, 25–46.

Fisher, I. (1933). The Debt-Defla­tion The­ory of Great Depres­sions, Econo­met­rica. 1: 337–57.

Fried­man, M. and A.J. Schwartz (1963). A Mon­e­tary His­tory of the United States 1867–1960. Prince­ton (NJ): Prince­ton Uni­ver­sity Press.

Gärt­ner, M. and F. Jung (2010). The Macro­eco­nom­ics of Finan­cial Crises: How Risk Pre­mi­ums and Liq­uid­ity Traps Affect Pol­icy Options, Inter­na­tional Advances in Eco­nomic Research. 17: 12–27.

Hicks, J.R. (1935). Wages and Inter­est: The Dynamic Prob­lem, The Eco­nomic Jour­nal. 45: 456–68.

Hicks, J.R. (1937). Mr. Keynes and the “Clas­sics”; a Sug­gested Inter­pre­ta­tion, Econo­met­rica. 5: 147–59.

Hicks, J.R. (1981). Is-Lm: An Expla­na­tion, Jour­nal of Post Key­ne­sian Eco­nom­ics. 3: 139–54.

Holmes, A.R. (1969). Oper­a­tional Con­straints on the Sta­bi­liza­tion of Money Sup­ply Growth, in: F.E. Mor­ris (ed.), Con­trol­ling Mon­e­tary Aggre­gates. Nan­tucket Island: The Fed­eral Reserve Bank of Boston: 65–77.

Ire­land, P.N. (2011). A New Key­ne­sian Per­spec­tive on the Great Reces­sion, Jour­nal of Money, Credit, and Bank­ing. 43: 31–54.

Keen, S. (1995). Finance and Eco­nomic Break­down: Mod­el­ing Minsky’s ‘Finan­cial Insta­bil­ity Hypoth­e­sis’, Jour­nal of Post Key­ne­sian Eco­nom­ics. 17: 607–35.

Keen, S. (2000). The Non­lin­ear Eco­nom­ics of Debt Defla­tion, in: W.A. Bar­nett, C. Chiarella, S. Keen, R. Marks, and H. Schn­abl (eds.), Com­merce, Com­plex­ity, and Evo­lu­tion: Top­ics in Eco­nom­ics, Finance, Mar­ket­ing, and Man­age­ment: Pro­ceed­ings of the Twelfth Inter­na­tional Sym­po­sium in Eco­nomic The­ory and Econo­met­rics. New York: Cam­bridge Uni­ver­sity Press: 83–110.

Keen, S. (2008). Keynes’s ‘Revolv­ing Fund of Finance’ and Trans­ac­tions in the Cir­cuit, in: R. Wray and M. Forstater (eds.) Keynes and Macro­eco­nom­ics after 70 Years. Chel­tenham: Edward Elgar: 259–78.

Keen, S. (2011a). Debunk­ing Eco­nom­ics: The Naked Emperor Dethroned?. Lon­don: Zed Books.

Keen, S. (2011b). A Mon­e­tary Min­sky Model of the Great Mod­er­a­tion and the Great Reces­sion, Jour­nal of Eco­nomic Behav­ior & Orga­ni­za­tion. Forth­com­ing.

Keynes, J.M. (1936). The Gen­eral The­ory of Employ­ment, Inter­est and Money. Lon­don: Macmil­lan.

Kyd­land, F.E. and E.C. Prescott (1990). Busi­ness Cycles: Real Facts and a Mon­e­tary Myth, Fed­eral Reserve Bank of Min­neapo­lis Quar­terly Review. 14: 3–18.

Lucas, R.E., Jr. (2003). Macro­eco­nomic Pri­or­i­ties, Amer­i­can Eco­nomic Review. 93: 1–14.

McK­ib­bin, W.J. and A. Stoeckel (2009). Mod­el­ling the Global Finan­cial Cri­sis, Oxford Review of Eco­nomic Pol­icy. 25: 581–607.

Min­sky, H.P. (1982). Can “It” Hap­pen Again?: Essays on Insta­bil­ity and Finance. Armonk, NY: M.E. Sharpe.

Moore, B.J. (1979). The Endoge­nous Money Stock, Jour­nal of Post Key­ne­sian Eco­nom­ics. 2: 49–70.

Moore, B.J. (1983). Unpack­ing the Post Key­ne­sian Black Box: Bank Lend­ing and the Money Sup­ply, Jour­nal of Post Key­ne­sian Eco­nom­ics. 5: 537–56.

Moore, B.J. (1988). The Endoge­nous Money Sup­ply, Jour­nal of Post Key­ne­sian Eco­nom­ics. 10: 372–85.

Moore, B.J. (2001). Some Reflec­tions on Endoge­nous Money, in: L.-P. Rochon and M. Ver­nengo (eds.), Credit, Inter­est Rates and the Open Econ­omy: Essays on Hor­i­zon­tal­ism. Edward Elgar: Chel­tenham: 11–30.

Rol­nick, A.J. (2010). Inter­view with Thomas Sar­gent, The Region. 24: 26–39.

Schum­peter, J.A. (1934). The The­ory of Eco­nomic Devel­op­ment: An Inquiry into Prof­its, Cap­i­tal, Credit, Inter­est and the Busi­ness Cycle. Cam­bridge, MA: Har­vard Uni­ver­sity Press.

Solow, R.M. (2003). Dumb and Dumber in Macro­eco­nom­ics, Festschrift for Joe Stiglitz. Colum­bia Uni­ver­sity:

Solow, R.M. (2001). From Neo­clas­si­cal Growth The­ory to New Clas­si­cal Macro­eco­nom­ics, in: J. H. Drèze (ed.), Advances in Macro­eco­nomic The­ory. New York: Pal­grave.

Solow, R.M. (2008). The State of Macro­eco­nom­ics, The Jour­nal of Eco­nomic Per­spec­tives. 22: 243–46.

About Steve Keen

I am Professor of Economics and Head of Economics, History and Politics at Kingston University London, and a long time critic of conventional economic thought. As well as attacking mainstream thought in Debunking Economics, I am also developing an alternative dynamic approach to economic modelling. The key issue I am tackling here is the prospect for a debt-deflation on the back of the enormous private debts accumulated globally, and our very low rate of inflation.
Bookmark the permalink.
  • Bhaskara II

    To Jh and Steve:

    Assem­bling the Japan­ese credit data points across spread sheet pages can be done in the fol­low­ing ways.

    I’m sure that If you are good at “pivot tables” that would be an easy way to get the data from across sheets. They take a few hours to get used to but every one who uses them loves them immensely. Pivot tables can be instan­ta­neously set up in 30 sec­onds, and mod­i­fied in 10 sec­onds. They can also do cal­cu­la­tions such as sum and aver­age on the data. Pivot tables are very fast to set up once the learn­ing curve is sur­mounted. There are pivot table lessons on line. This might be the best way in the future. A busi­ness or CS stu­dent would prob­a­bly be able to show you how they work.

    Shown below is an other way, using for­mu­las con­tain­ing look-up and text func­tions. This way can be expressed in text and works on almost all spread sheet pro­grams. If copied exactly this might be a quick way to get the data. 

    Here is an exam­ple spread­sheet to make a col­umn of cells from data that is dis­trib­uted across sep­a­rate sheets named Sheet1, Sheet2, …, Sheet6. 

    The func­tions used are:

    Address Forms a string (text) name of address from
    row, col­umn, sheet , and file (not used here).
    Indi­rect Fetches data (indi­rectly) using the address string.
    Con­cate­nate Con­cate­nate assem­bles a string (text), ie “Sheet1”
    from “Sheet” and 1.

    The address func­tion instruc­tions show that it is also pos­si­ble is to to do this across “work­books” or spread­sheet files. 

    ###### Start Exam­ple instruc­tions:

    The fol­low­ing is all entered on sheet1 but will ref­er­ence other sheets.

    Cells with sheet num­bers:
    cell con­tents of cell
    a39 1
    a40 2
    a41 3
    a41 4
    a42 5
    a44 6 (Num­bers past 1 could be cal­cu­lated.)
    b1 ’dat­a­point’

    This for­mula returns value in the cell of row 1, col­umn 2, and “Sheet1”
    cell: b39
    for­mula: =INDIRECT(ADDRESS(1,2„,CONCATENATE(“Sheet”,A39)))

    Copy the for­mula to cells below B39 for the other sheet num­bers. A39 in the for­mula should change to ref­er­ence the other num­bers when copied to other cells.

    In this case the data wanted is in row 1, col­umn 2, i.e. cell B1 and in Sheet1 thru Sheet6.

    You can fill b1 cells on the other sheets with data if you like.

    The “„,” leaves two of the (address) func­tion para­me­ters blank. 

    ###### END Exam­ple:

    If on sets this up from scratch, rather than copy­ing the exam­ple given (quick­est), it is best to use a sin­gle func­tion per col­umn in order to get all the para­me­ters cor­rect. When all the func­tions work, they could be used with mul­ti­ple columns or assem­bled into a sin­gle for­mula. This took me about 1/5 to 1.5 hours from scratch using the instruc­tions.

    If you use a dif­fer­ent spread­sheet pro­gram the func­tion names are prob­a­bly the same or sim­i­lar. In that case, look at the man­ual, help, online exam­ples, and/or use a func­tion but­ton where you fill in the blanks and it makes the for­mula.

    In the address instruc­tions below they use “[Book1] ” to ref­er­ence work­book1. I believe that is how you can fetch data across files. In that case you would add para­me­ters to the con­cate­nate func­tion and add an other col­umn that has the book num­bers in it.

    Instruc­tions Links for the excel func­tions used above are as fol­lows:–01-30-excel-2010-concatenate-text-strings.htm

  • cliffy


    Is social jus­tice para­mount over prop­erty rights — yes.

    For a cause to qual­ify as a social jus­tice cause, how­ever, it must be based in an abil­ity to secure an equi­table out­come based on the pro­pri­etor of the cause what Aus­tralians would term “hav­ing a go” in self procur­ing the con­di­tion of which the indi­vid­ual is cur­rently denied.

    What’s inter­est­ing with the notion of lend­ing to the asset-less, not-with­stand­ing the fact that your descrip­tion of money mak­ing money as an injus­tice amongst scarce money is a strong point, and a point I accept, and to be weighed up in terms of jus­tice, is that:

    1. To some extent hav­ing assets can be a track record of an abil­ity to man­age that which will take your money over a longer period of time. If debt is to held back from the Min­sky / Schum­peter bound­ary then it will be pro­duc­tive due to there being less of it?

    2. Isn’t this what the US just did with things like NINJA loans?

  • cliffy


    then it will *need to* be pro­duc­tive

  • cliffy

    Mahaish, RJ,

    Aren’t your views uni­fied by accept­ing that yes account­ing prac­tice and rules and indeed reg­u­la­tion pinned to account­ing rules around deci­sions lim­it­ing to deci­sions to an extent, but that equally if a polit­i­cally ade­quate body deter­mines that the US will just keep adding quan­tum to the account it holds in the data­bases that feeds into trans­ac­tional sys­tems in the US and beyond then they can phys­i­cally do it with­out accept­ing it to be in the nature of a debt and with­out hav­ing to deduct bal­ances from other records of that same data­base?

    Does that accom­mo­date both your posi­tions?

  • cliffy


    pinned to account­ing rules around deci­sions lim­it­ing to deci­sions to an extent

    should read:

    pinned to account­ing rules around deci­sions limit deci­sions to an extent

  • alain­ton


    scott ful­willers treat­ment of bonds — and MMTs in gen­eral — is flawed. 

    Bonds are far less infla­tion­ary than direct cred­it­ing of bank accounts. 

    Buy­ing a bond trans­fers pri­vate sec­tor sur­plus to pay­ing down a gov­ern­ment debt, it can­cels the debt by that amount destroy­ing the (non coupon) money and is neu­tral on effec­tive demand and the money sup­ply.

    The pay­ment of the coupon does expand the money sup­ply — in the future — and if the econ­omy is expand­ing then tax rev­enues can afford the pay­ment of inter­est in favourable debt dynam­ics. As this is spent tax­a­tion it will not expand the money sup­ply and effec­tive demand so as the tax­a­tion with­draws money from cir­cu­la­tion and the spend­ing on the coupon puts it back in the pri­vate sec­tor, if you have to pay debt on debt, then the bor­row­ing to pay this in the domes­tic econ­omy can be defla­tion­ary. It is only infla­tion­ary if the econ­omy is not expand­ing and you have to bor­row from abroad . As bonds are used as a liq­uid­ity hedge against infla­tion it will be dif­fi­cult to sus­tain a risk free rate less than infla­tion.

    Direct cred­it­ing of bank account does not with­draw pri­vate sec­tor sur­pluses, every cent of the pay­ment expands the domes­tic money sup­ply. If this is used for invest­ment need not be auto­mat­i­cally infla­tion­ary, but if used to sustina rev­enue expen­di­ture will be.

  • Lyon­wiss

    @ Alain­ton Decem­ber 12, 2011 at 10:28 pm

    Agree. I’m quite sick of read­ing every­where on the inter­net that hav­ing a sov­er­eign cur­rency and a print­ing press, solves all eco­nom­ics prob­lems. Gov­ern­ment debt is no prob­lem, even desir­able, blah, blah, blah — the blind, with­out a clue, lead­ing the sheep, with­out a brain. Sim­plify a prob­lem enough, a solu­tion emerges.

  • alain­ton

    New bank of Eng­land report sug­gest­ing the inter­na­tional finan­cial sys­tem works far less well than under the Bret­ton Woods sys­tem and sug­gest­ing a tax on cap­i­tal flows — A tar­iff ! knock me down with a feather

    Also debunks the ABC the­ory that low inter­est rates were the cause of the finan­cial cri­sis.

    Still to my mind doesn’t explain why cap­i­tal has started to flow ‘uphill’ — only the level of debt in the west can explain that. Sav­ing sur­plus in east = debt in west.

  • Matthew K

    Thank you Lyon­wiss and A for chip­ping in here. 

    Its one thing to per­son­ally ‘mis­take the map for the ter­ri­tory’ but I don’t think we should let it pass when being con­fi­dently proph­e­sied to oth­ers.

    Steve’s blog is built upon the empir­i­cal while MMT is the appli­ca­tion of the­ory at the expense of the empir­i­cal.–1

  • Matthew K

    Alain­ton — ‘Still to my mind doesn’t explain why cap­i­tal has started to flow ‘uphill’ – only the level of debt in the west can explain that. Sav­ing sur­plus in east = debt in west.’ 

    Can you clar­ify what you mean by ‘uphill’?

  • alain­ton


    Can you clar­ify what you mean by ‘uphill’?’

    ref­er­ence to the BOE paper — under nor­mal cir­cum­stances cap­i­tal should flow ‘down­hill’ from cap­i­tal rich/labour poor economies to the oppo­site — because of the dimin­ish­ing returns to cap­i­tal — since at least the 1980s we have seen the reverse process with cap­i­tal flow­ing in the other direc­tion — ‘uphill’ accord­ing to neo­clas­si­cal the­ory this should not be hap­pen­ing, indeed Robert Lucas wrote an arti­cle

    Lucas, Robert, 1990, “Why Doesn’t Cap­i­tal Flow from Rich to Poor Coun­tries?” Amer­i­can Eco­nomic Review, Vol. 80 (May), pp. 92–96.

    stat­ing the flows were not those pre­dicted by the­ory, indeed it has been called the ‘lucas para­dox’ — — although the real lucas para­dox is why he is taken seri­ously at all.

    see this IMF paper
    Or this much cited

  • cliffy


    Surely the appli­ca­tion of those big trade sur­pluses pil­ing up in China qual­ify as Cap­i­tal in pure eco­nomic thought?

  • mahaish

    its false dichotomy whether the assett is held as deposits or bonds

    bonds can be liq­ui­dated , lever­aged and col­la­toralised in a blink of an eye 


    your line of think­ing is clas­sic igbc, and its a flawed frame­work

  • mahaish

    i think the impor­tant point is

    its the total level of the fis­cal deficit that may be infla­tion­ary,

    not the liq­uid­ity and port­fo­lio com­pos­tion of the deficit that the cen­tral bank tries to impose,

  • RJ

    Buy­ing a bond trans­fers pri­vate sec­tor sur­plus to pay­ing down a gov­ern­ment debt”

    How can buy­ing a bond pay down Govt debt.

    It clearly does not so I’m unsure what you mean by the above state­ment

    And what does this mean. Do you mean inter­est on debt? And bor­row­ing by who. And bor­row­ing (or increased Govt debt) increases the money sup­ply or finan­cial assets so how can it be defla­tion­ary?

    if you have to pay debt on debt, then the bor­row­ing to pay this in the domes­tic econ­omy can be defla­tion­ary.

  • RJ

    Gov­ern­ment debt is no prob­lem, even desir­able, blah, blah, blah ”

    It is desir­able and essen­tial to pro­vide the net non Govt finan­cial asset required for retire­ment sav­ings.

    Igno­rance of this easy to prove point is the cause of many of our cur­rent eco­nomic prob­lems

  • @Steve Keen Decem­ber 11, 2011 at 6:46 am

    For exam­ple, id love to use the Japan­ese data, but their credit data is stored in a spread­sheet with one page per year! The work involved in import­ing that ver­sus load­ing a col­umn of data from a PRN file is just too much.

    What exactly does the PRN file for­mat look like? Can you pro­vide a link to such a file? I can covert HTML table data to excel and visa verso.

    From here,

    a PRN file is either Post­Script and PCL. I need to know how it is for­mat­ted as a string since it may be pos­si­ble to con­vert Excel to PRN but by non con­ven­tional means.

  • clive

    RE Japan­ese files. If they are late exel files you should be able to save as xml. Once they are in xml for­mat you should be able to do pretty well any­thing with them.

  • Lyon­wiss

    @ Alain­ton Decem­ber 13, 2011 at 7:21 am

    Econ­o­mists have a very stu­pid epis­te­mol­ogy, which is to start with a
    very sim­ple model, with unre­al­is­tic assump­tions (a la Mil­ton Fried­man). Then
    they find all sorts of para­doxes, anom­alies and conun­drums and then they write
    thou­sands of research papers dis­cussing why real­ity doesn’t fit their beau­ti­ful,
    sim­ple (aka ele­gant) and rig­or­ous model. Then they con­clude the prob­lem is
    real­ity, which must be changed! The ter­ri­tory has to be altered to fit the map!

    Uphill cap­i­tal flow? No, it’s impos­si­ble to defy the law of grav­ity. The data
    must be wrong. The way to pros­per­ity is to print more money, cre­ate more
    gov­ern­ment debt. Jobs are guar­an­teed to cre­ate wealth even dig­ging holes or
    break­ing win­dows. What sov­er­eign debt cri­sis? The Euro­peans and Amer­i­cans are only imag­in­ing things. Only clear think­ing, detached econ­o­mists sit­ting in their arm­chairs can cre­ate the per­fect econ­omy.

    How about start­ing with the real world, with a list of obser­va­tions you want to explain, before you build your model?

  • alain­ton


    If there is gov­ern­ment debt (neg­a­tive bal­ance sheet) then if that debt is not mon­e­tised then their will have to be a sec­toral pay­ment from the pri­vate sec­tor (in a closed econ­omy) or net trans­fer from abroad. What mat­ters here is the neg­a­tive bal­ance sheet — out­stand­ing lia­bil­i­ties includ­ing pay­ment of inter­est from pre­vi­ous loans being greater than assets from taxes etc. 

    I was foggy on the issue of deflation/inflation. If you are tax­ing at home/cutting spend­ing and pay­ing down loans it is con­trac­tionary and defla­tion­ary in cur­rent cir­cum­stances. If you are issu­ing bonds to pay the coupon on pre­vi­ously issued bonds it is expan­sion­ary and can be infla­tion­ary if there is no excess capac­ity. The is just the credit impulse/accelerator again.


    Cant see any­thing of igbc here — all expressed in very post Key­ne­sian terms of sec­toral bal­ances, credit impulse, taxes remov­ing money from cir­cu­la­tion and gov­ern­ment spend­ing putting it back in cir­cu­la­tion.

    Any asset/capital/bond can be liq­ui­dated — doesn’t mean they are straight cash equiv­a­lents, liq­uid­ity pref­er­ence, intertem­po­ral cap­i­tal struc­ture and Q mat­ters. They are only straight equiv­a­lents at zero inter­est rates. The rela­tion­ship between the two is crit­i­cal as firms need to opti­mise their cap­i­tal struc­ture to max­imise NPV — so it is a key deter­mi­nate of invest­ment.

  • koonyeow

    Title: Just Express­ing Myself

    From Lyon­wiss:

    Econ­o­mists have a very stu­pid epis­te­mol­ogy, which is to start with a very sim­ple model, with unre­al­is­tic assump­tions (a la Mil­ton Fried­man). Then they find all sorts of para­doxes, anom­alies and conun­drums and then they write thou­sands of research papers dis­cussing why real­ity doesn’t fit their beau­ti­ful, sim­ple (aka ele­gant) and rig­or­ous model. Then they con­clude the prob­lem is real­ity, which must be changed! The ter­ri­tory has to be altered to fit the map!

    Uphill cap­i­tal flow? No, it’s impos­si­ble to defy the law of grav­ity. The data must be wrong. The way to pros­per­ity is to print more money, cre­ate more gov­ern­ment debt.

    Can I call that descrip­tion ‘Lec­tur­ing Birds On Fly­ing’?

    Jobs are guar­an­teed to cre­ate wealth even dig­ging holes or break­ing win­dows.

    Smells like Bas­tiat to me.

    Only clear think­ing, detached econ­o­mists sit­ting in their arm­chairs can cre­ate the per­fect econ­omy.

    I can be detached when I’m sit­ting in an arm­chair, but my think­ing dur­ing those arm­chair ses­sions is never clear. Per­haps that is why I’m not an (neo-clas­si­cal) econ­o­mist.

    How about start­ing with the real world, with a list of obser­va­tions you want to explain, before you build your model?

    Smells like how ‘evo­lu­tion by nat­ural selec­tion’ was dis­cov­ered.

  • Lyon­wiss

    @ Koonyeow Decem­ber 14, 2011 at 1:31 am

    You should have men­tioned the sub­stance of many posts in this thread: “What sov­er­eign debt cri­sis? The Euro­peans and Amer­i­cans are only imag­in­ing things.” Smells like MMT?

  • Matthew K

    AND here’s kyle bass on Japan–2011/#clip584882

    when you look at their bal­ance sheet and income state­ment… its the sin­gle most obvi­ous thing i’ve seen in my adult life’

    alain­ton — thanks for the links, read most of them! could we argue that that uphill trend has reversed in the sec­ond half of the last decade?

  • mahaish

    Any asset/capital/bond can be liq­ui­dated – doesn’t mean they are straight cash equiv­a­lents, liq­uid­ity pref­er­ence, intertem­po­ral cap­i­tal struc­ture and Q mat­ters. They are only straight equiv­a­lents at zero inter­est rates. The rela­tion­ship between the two is crit­i­cal as firms need to opti­mise their cap­i­tal struc­ture to max­imise NPV – so it is a key deter­mi­nate of invest­ment”

    hi ala­ianton

    well i have to dis­agree regard­ing the liq­uid­ity aspect, given how much trade goes on in these mar­kets, and im not so sure about the propo­si­tion that they are staight equiv­i­lants at zero rates,

    since short term trea­suries are a proxy for the inter­bank rate which is a equiv­i­lant to the cash rate.

    the net spend­ing by the gov­ern­ment is still left in cir­cu­la­tion, and i think that is the key point, in terms of any infla­tion­ary effect.

    admit­tedly port­fo­lio com­posi­ton of that net spend­ing may cause infla­tion in cer­tain finan­cial assett classes that are the sub­ject of cen­tral bank manip­u­la­tion,

    the cen­tral bank does this reg­u­larly by try­ing to increase the face value of secu­ri­ties by restrict­ing sup­ply in order to manip­u­late the yield curve.

    but i take your point about igbc, i may have mis­con­strued what you have said and my apolo­gies.

  • Lyon­wiss

    The lat­est US esti­mate of MMT in action: