Mod­el­ing Finan­cial Insta­bil­ity

Flattr this!

This paper will be pub­lished in a forth­com­ing book on the cri­sis edited by Malliaris, Shaw and She­frin. In what fol­lows, I derive a cor­rected for­mula for the role of the change in debt in aggre­gate demand, which is that ex-post aggre­gate demand equals ex-ante income plus the cir­cu­la­tion of new debt, where the lat­ter term is the veloc­ity of money times the ex-post cre­ation of new debt.

The PDF is avail­able here: Keen2014­Mod­el­ing­Fi­nan­cialIn­sta­bil­ity. The Min­sky mod­els used in this paper are here in a ZIP file. The lat­est ver­sion of Min­sky can be down­loaded from here.

  • Introduction

Lit­er­ally no-one dis­putes that the finan­cial sec­tor was the cause of the post-2007 eco­nomic cri­sis: dis­pu­ta­tion instead cen­ters on the causal mech­a­nisms. I fol­low Fisher (Fisher 1933) and Min­sky (Min­sky 1980) in assign­ing key roles to the growth and con­trac­tion of aggre­gate pri­vate debt (Keen 1995; Keen 2000), but this per­spec­tive is rejected by New Key­ne­sian econ­o­mists on the a pri­ori basis that pri­vate debts are “pure redis­tri­b­u­tions” that “should have no sig­nif­i­cant macro-eco­nomic effects” (Bernanke 2000p. 24), and as a corol­lary to the oft-repeated tru­ism that “one person’s debt is another person’s asset” (Krug­man 2012c, p. 43).

My analy­sis also fol­lows the Post Key­ne­sian tra­di­tion of endoge­nous money (Moore 1979; Moore 1983) in see­ing the bank­ing sec­tor as an essen­tial com­po­nent of the macro­econ­omy, yet this is also dis­missed by New Key­ne­sian econ­o­mists on the grounds that banks are merely a spe­cial­ized form of finan­cial inter­me­di­ary (Krug­man 2012a; Krug­man 2012b; Krug­man 2013a; Sum­ner 2013; Tobin 1963), all of which can be safely ignored in macro­eco­nomic mod­els. When banks are intro­duced in New Key­ne­sian mod­els, they func­tion not as loan orig­i­na­tors but effec­tively as bro­kers between savers and bor­row­ers (Eggerts­son and Krug­man 2012b, pp. 21–22).

In response, authors in the Post Key­ne­sian and Endoge­nous Money tra­di­tions express exas­per­a­tion that New Key­ne­sian authors ignore credit cre­ation and the account­ing mechan­ics of bank lend­ing (Full­wiler 2012; Roche 2013), as laid out in numer­ous Cen­tral Bank pub­li­ca­tions (Car­pen­ter and Demi­ralp 2010; ECB 2012; Holmes 1969; Keis­ter and McAn­drews 2009).

Given the key pub­lic pol­icy role of eco­nom­ics, and the acknowl­edged fail­ure of Neo­clas­si­cal mod­els in gen­eral to antic­i­pate the finan­cial cri­sis (Beze­mer 2009; Blan­chard 2009; Blan­chard, et al. 2010; OECD 2007), the exis­tence within aca­d­e­mic eco­nom­ics of two dia­met­ri­cally opposed per­spec­tives which fail to com­mu­ni­cate is a dis­ser­vice to the pub­lic.

In this paper I attempt to con­clu­sively deter­mine whether aggre­gate pri­vate debt and banks mat­ter in macro­eco­nom­ics by putting the two rival mod­els of lending—Loanable Funds and Endoge­nous Money—on a com­mon foot­ing. Using the dynamic Open Source mon­e­tary mod­el­ing pro­gram Min­sky, I firstly put the New Key­ne­sian model of bank­ing in Eggerts­son & Krug­man 2012b into a strictly mon­e­tary model and I show that, if the struc­ture of lend­ing in this model accu­rately char­ac­ter­izes actual lend­ing, then the Neo­clas­si­cal per­spec­tive that aggre­gate debt is unim­por­tant, and that banks can safely be ignored in macro­eco­nom­ics, is cor­rect. I then mod­ify this model to match the Post Key­ne­sian per­spec­tive on the struc­ture of lend­ing, and show that in this struc­ture, changes in the aggre­gate level of pri­vate debt have a direct impact upon aggre­gate demand, and banks there­fore play a cru­cial role in macro­eco­nom­ics.

  1. Loanable Funds vs Endogenous Money

The Neo­clas­si­cal model of “Loan­able Funds” and the Post Key­ne­sian con­cept of “Endoge­nous Money” con­sti­tute the polar oppo­sites on the nature and sig­nif­i­cance of banks, debt and money in macro­eco­nom­ics. Both mod­els por­tray the money sup­ply as vari­able, and hence in one sense endoge­nous, though by very dif­fer­ent mech­a­nisms and to very dif­fer­ent degrees (Pal­ley 2013, p. 411). In the Loan­able Funds tra­di­tion, banks func­tion as “mere inter­me­di­aries” (Graziani 1989, p. 8) between savers and bor­row­ers, pri­vate debts are “pure redis­tri­b­u­tions” that “should have no sig­nif­i­cant macro-eco­nomic effects” (Bernanke 2000, p. 24), and banks, debt and money can be and are ignored in canon­i­cal macro­eco­nomic mod­els (Smets and Wouters 2007; Wood­ford 2009). In the Endoge­nous Money tra­di­tion, banks are cru­cial to macro­eco­nom­ics because they cre­ate money by cre­at­ing debt (Holmes 1969; Moore 1979), but no con­sen­sus has yet emerged on how to rep­re­sent this phe­nom­e­non in Post Key­ne­sian macro­eco­nomic mod­els (Pal­ley 1991; Pal­ley 2002).

There is lit­tle com­mu­ni­ca­tion between the two approaches, with authors in the Loan­able Funds tra­di­tion fre­quently derid­ing those in the Endoge­nous Money camp (Krug­man 2012a; Krug­man 2012b; Krug­man 2012e; Krug­man 2012f), and dis­miss­ing the propo­si­tion that banks must be included in macro­eco­nom­ics (Krug­man 2012a; Krug­man 2012b; Krug­man 2012d; but see Rowe 2013; Sum­ner 2013).

This dis­pute can be resolved by an appeal to the Occam’s Razor prin­ci­ple that unless a more com­plex model makes dif­fer­ent and bet­ter pre­dic­tions than a less com­plex one, the sim­pler should be pre­ferred. There­fore, unless bank lend­ing nec­es­sar­ily affects vital macro­eco­nomic aggre­gates in a sig­nif­i­cant man­ner, then even though the “loans cre­ate deposits” account­ing per­spec­tive of Endoge­nous Money is tech­ni­cally cor­rect (Car­ney 2012; ECB 2012; Holmes 1969)—as even Paul Krug­man has con­ceded (Krug­man 2013a)— the Loan­able Funds approach is jus­ti­fied, and banks should be excluded from macro­eco­nom­ics. Con­versely, if bank lend­ing nec­es­sar­ily affects macro­eco­nomic aggre­gates, then banks, debt and the endo­gene­ity of the money sup­ply are inte­gral to macro­eco­nom­ics, and mod­els that exclude them are not mod­els of a cap­i­tal­ist econ­omy.

  1. A monetary model of Loanable Funds

Eggerts­son and Krug­man note that the vast major­ity of main­stream eco­nomic mod­els ignore debt:

If there is a sin­gle word that appears most fre­quently in dis­cus­sions of the eco­nomic prob­lems now afflict­ing both the United States and Europe, that word is surely debt… one might have expected debt to be at the heart of most main­stream macro­eco­nomic models—especially the analy­sis of mon­e­tary and fis­cal pol­icy. Per­haps some­what sur­pris­ingly, how­ever, it is quite com­mon to abstract alto­gether from this fea­ture of the econ­omy. Even econ­o­mists try­ing to ana­lyze the prob­lems of mon­e­tary and fis­cal pol­icy at the zero lower bound—and yes, that includes the present authors (see Krug­man 1998, Eggerts­son and Wood­ford 2003)—have often adopted rep­re­sen­ta­tive agent mod­els in which every­one is alike and the shock that pushes the econ­omy into a sit­u­a­tion in which even a zero inter­est rate is not low enough takes the form of a shift in everyone’s pref­er­ences. (Eggerts­son and Krug­man 2012a, pp. 1469–71)

In order to intro­duce debt into a New Key­ne­sian two-period model, Eggerts­son and Krug­man divided agents into two groups who “dif­fer only in their rates of time pref­er­ence”: “patient agents” and “impa­tient agents” where the lat­ter have a higher rate of time pref­er­ence than the for­mer, so that “In that case, ”impa­tient” indi­vid­u­als will bor­row from ”patient” indi­vid­u­als.” (Eggerts­son and Krug­man 2012ap. 1474). Debt was explic­itly mod­eled through­out this paper, and bank­ing was intro­duced in the Appen­dix (Eggerts­son and Krug­man 2012b) as an inter­me­di­at­ing func­tion between depos­i­tors and bor­row­ers, where bor­row­ing by impa­tient agents was strictly for invest­ment.

The authors describe their model as a “just the stan­dard New Key­ne­sian model”, with one twist, in that the nat­ural rate of inter­est, which is nor­mally an exoge­nous para­me­ter in the IS equa­tion, is instead endoge­nous with bor­row­ers’ debt being one of its para­me­ters. There­fore the level of pri­vate debt plays a macro­eco­nomic role:

we need to fig­ure out the evo­lu­tion of debt of the “bor­row­ers” to fig­ure out the nat­ural rate of inter­est. In par­tic­u­lar we see that if … the econ­omy is “over­lever­aged” … it is easy to get endoge­nously neg­a­tive nat­ural rate of inter­est. (Eggerts­son and Krug­man 2012b, p. 24)

The New Key­ne­sian and “Liq­uid­ity Trap” aspects of this model (on which see Solow 2003; Solow 2008) are tan­gen­tial to the topic of this paper, which is a strictly struc­tural one: does bank lending—as opposed to lend­ing by non-bank agents to each other—significantly alter the macro­dy­nam­ics of the econ­omy? To con­sider this ques­tion, I ren­der the Loan­able Funds aspects of Eggerts­son and Krug­man 2012b in a strictly mon­e­tary form in a Min­sky model.

Min­sky is a sys­tem dynam­ics pro­gram which gen­er­ates dynamic mod­els of finan­cial flows from dou­ble-entry book­keep­ing tables (called “God­ley Tables” in the pro­gram), in which the columns rep­re­sent bank accounts and the rows are trans­ac­tions between accounts. The sam­ple model shown in Fig­ure 1 gen­er­ates the dynamic equa­tions shown in Equa­tion (more details on Min­sky are given in the Appen­dix).

Fig­ure 1: Sam­ple God­ley Table and bank­ing icon in Min­sky

The Loan­able Funds fea­tures of Eggerts­son and Krug­man (2012b) are:

  • that deposits by the “patient agents” enable loans to “impa­tient agents”; and
  • that banks inter­me­di­ate between saver and bor­rower and profit by an inter­me­di­a­tion fee, but oth­er­wise play no role in lend­ing.

The Min­sky model shown in Fig­ure 2 repli­cates these fea­tures using the bank accounts of four sep­a­rate enti­ties: the con­sump­tion goods sec­tor (with deposit account DepCons) which is the lender in (Eggerts­son and Krug­man 2012b); the invest­ment goods sec­tor (with account DepInv) which is the bor­rower; Work­ers (with account Work­ers) who are employed by both the Con­sump­tion Sec­tor and the Invest­ment Sec­tor; and
the Bank­ing sec­tor (with the Asset account Reserves and equity account BankersNW) which inter­me­di­ates the loans from
the Con­sump­tion Sec­tor to the Invest­ment Sec­tor, and charges a fee for doing so. Each sec­tor main­tains a finan­cial table show­ing the flows into and out of its accounts, and cal­cu­lates its net worth as a result as the dif­fer­ence between the value of its assets and lia­bil­i­ties (account BankersNW for the bank­ing sec­tor).

Fig­ure 2: Loan­able Funds model—a 4 account view of Loan­able Funds gen­er­ated in Min­sky

Table 1 shows this finan­cial sys­tem from the bank­ing sector’s per­spec­tive, and Table 2 shows it from the per­spec­tive of the lender, the Con­sump­tion Sec­tor. Fol­low­ing the con­ven­tions in Min­sky, assets are shown as pos­i­tive amounts, and lia­bil­i­ties and equity are shown as neg­a­tives, while the source of any finan­cial trans­ac­tion is shown as a pos­i­tive and its des­ti­na­tion as a neg­a­tive. All entries in the table rep­re­sent flows, and Min­sky auto­mat­i­cally gen­er­ates the result­ing sys­tem of dif­fer­en­tial equa­tions in LaTeX. The ten flows that define the model are all shown in the bank­ing sector’s table, and are respec­tively:

  1. The Con­sump­tion Sec­tor lends to the Invest­ment Sec­tor via the flow “Lend” from the account DepCons to the account DepInv;
  2. The Invest­ment sec­tor makes Inter­est pay­ments “Int” to the con­sump­tion sec­tor;
  3. The Bank­ing Sec­tor charges the Con­sump­tion Sec­tor an inter­me­di­a­tion fee “IntFee”;
  4. The invest­ment Sec­tor makes debt repay­ments to the Con­sump­tion Sec­tor (“Repay”);
  5. The Con­sump­tion Sec­tor hires Work­ers via the flow “WagesC”;
  6. The invest­ment Sec­tor hires Work­ers via the flow “WagesI”;
  7. The Invest­ment Sec­tor pur­chases con­sump­tion goods (“ConsI”);
  8. The Con­sump­tion Sec­tor pur­chases invest­ment goods (“ConsC”);
  9. Work­ers pur­chase con­sumer goods (“ConsW”); and
  10. Bankers pur­chase con­sumer goods (“ConsB”);

Table 1: Loan­able Funds model from the Bank­ing Sector’s per­spec­tive

Bank­ing Sec­tor




Flows Accounts Reserves DepCons DepInv Work­ers BankersNW
1 Lend­ing Lend –Lend
2 Inter­est Pay­ments –Int Int
3 Bank Inter­me­di­a­tion Fee IntFee –IntFee
4 Debt Repay­ment –Repay Repay
5 Hire work­ers (Cons) WagesC –WagesC
6 Hire work­ers (Inv) WagesI –WagesI
7 Inter­sec­toral pur­chases by Inv –ConsI ConsI
8 Inter­sec­toral pur­chases by Cons ConsC –ConsC
9 Work­ers con­sump­tion –ConsW ConsW
10 Bankers con­sump­tion –ConsB ConsB


Lend­ing from the con­sump­tion to the invest­ment sec­tor is recorded in the account Loans, which is an asset of the con­sump­tion sec­tor as shown in its finan­cial account (see Table 2; it also appears as a lia­bil­ity of the Invest­ment Sec­tor in its table of accounts; Table 2 also dis­plays the dynam­ics of the Con­sump­tion Sector’s net worth in the col­umn “ConsNW”).

Table 2: Loan­able Funds model from the Con­sump­tion Sector’s per­spec­tive

Con­sump­tion Sec­tor Assets Equity
Flows Accounts DepCons Loans ConsNW
1 Lend­ing –Lend Lend
2 Inter­est Pay­ments Int –Int
3 Bank Inter­me­di­a­tion Fee –IntFee IntFee
4 Debt Repay­ment Repay –Repay
5 Hire work­ers (Cons) –WagesC WagesC
6 Inter­sec­toral pur­chases by Inv ConsI –ConsI
7 Inter­sec­toral pur­chases by Cons –ConsC ConsC
8 Work­ers con­sump­tion ConsW –ConsW
9 Bankers con­sump­tion ConsB –ConsB


Since (for the sake of sim­plic­ity) hold­ings of cash are ignored in this model, money is the sum of the amounts in the four deposit accounts DepCons, DepInv, Work­ers, and BankersNW shown in Table 1, while debt is the amount in the account Loans shown in Table 2. Equa­tion shows the equa­tions for the dynam­ics of money and debt in the model, with the first 4 equa­tions derived from Table 1 show­ing the dynam­ics of money in the sys­tem while the final equa­tion, derived from Table 2, shows the dynam­ics of debt.

Defin­ing money M as the sum of the first four accounts, it is obvi­ous that the change in the amount of money is zero:

There­fore the amount of money—which for con­ve­nience we can treat this as hav­ing been cre­ated by gov­ern­ment fiat, with­out need­ing to spec­ify a gov­ern­ment sec­tor in the model—remains con­stant:

With­out hav­ing to define a full eco­nomic model, we can now spec­ify aggre­gate demand AD as being equiv­a­lent to the turnover of the money in the econ­omy, using the veloc­ity of money v (see Fig­ure 3 and Equa­tion ).

Fig­ure 3: Veloc­ity of M2 money stock in the USA 1960–2013

As is well known, con­trary to Mil­ton Friedman’s claims (Fried­man 1948; Fried­man 1959; Fried­man 1969; Fried­man and Schwartz 1963), the veloc­ity of money is not a constant—“it is also appar­ent that money veloc­i­ties are pro­cycli­cal and quite volatile” (Kyd­land and Prescott 1990, p. 14). How­ever the iden­tity that can be used in this sim­ple model to map from the money stock to the level of aggre­gate demand.

Using the sub­script LM to indi­cate that this is aggre­gate demand in a Loan­able Funds model, we have that aggre­gate demand at time t is the veloc­ity of money times the stock of money at that time:

Aggre­gate demand across any defined time period t2–t1 will there­fore be this instan­ta­neous flow times the time period itself:

Finally, using D for brevity in place of Loans in Equa­tion , it is obvi­ous that there is no link between the dynam­ics of debt and either the stock or the turnover of money, and there­fore there is no direct rela­tion between pri­vate debt and aggre­gate demand. The amount of money in cir­cu­la­tion remains con­stant:

Given the absence of a rela­tion­ship between lend­ing and the money sup­ply, the amount of debt in exis­tence can rise or fall sub­stan­tially with only a minor impact on macro­eco­nomic activ­ity via related changes in the veloc­ity of money:

  1. A monetary model of Endogenous Money

This struc­tural model of Loan­able Funds shown in Fig­ure 2 is con­verted into a model of Endoge­nous Money by three sim­ple changes:

  • Loans are shifted from the assets of the con­sump­tion sec­tor to the assets of the bank­ing sec­tor;
  • Inter­est pay­ments are trans­ferred to the equity account of the bank­ing sec­tor, BankersNW; and
  • Since banks are loan orig­i­na­tors in this model and receive inter­est pay­ments, the inter­me­di­a­tion fee is deleted.

This revised model is shown in Fig­ure 4 and Table 3. The changes between the Loan­able Funds model in Table 1 and the Endoge­nous Money model of Table 3 all occur in the first four rows, with the row for an inter­me­di­a­tion fee deleted, and loca­tions of the flows Lend, Int and Repay altered as indi­cated by the arrows. The two tables are oth­er­wise iden­ti­cal.

Fig­ure 4: Endoge­nous Money model in Min­sky

Table 3: Endoge­nous Money model from the bank­ing sector’s per­spec­tive

Bank­ing Sec­tor




Flows Accounts Reserves Loans DepCons DepInv Work­ers BankersNW
1 Lend­ing Lend  <– –Lend
2 Inter­est Pay­ments –> Int –Int
3 Debt Repay­ment –Repay <– Repay
4 Hire work­ers (Cons) WagesC –WagesC
5 Hire work­ers (Inv) WagesI –WagesI
6 Inter­sec­toral pur­chases by Inv –ConsI ConsI
7 Inter­sec­toral pur­chases by Cons ConsC –ConsC
8 Work­ers con­sump­tion –ConsW ConsW
9 Bankers con­sump­tion –ConsB ConsB


The money and debt equa­tions of this model are:

Despite the sim­plic­ity of the changes needed to move from Loan­able Funds to Endoge­nous Money, the dynam­ics of money are now pro­foundly dif­fer­ent. The rate of change of money is pre­cisely equal to the rate of change of debt:

The stock of money in the econ­omy is there­fore the sum of the ini­tial level of money in exis­tence, plus the new money cre­ated by the exten­sion of new loans from the bank­ing sec­tor to the invest­ment sec­tor. Assum­ing for con­ve­nience that D(0)=0, this yields:

Using the sub­script EM to indi­cate that this is an Endoge­nous Money model, aggre­gate demand is there­fore

Aggre­gate demand dur­ing some given time period t2–t1 is there­fore:

We can now com­pare the sym­bolic mea­sure of nom­i­nal aggre­gate demand in an Endoge­nous Money model with its coun­ter­part in a Loan­able Funds model (the numer­i­cal val­ues of veloc­ity, demand and debt will clearly dif­fer sub­stan­tially, as the sim­u­la­tions in Sec­tion 6 illus­trate) to iden­tify the sub­stan­tive dif­fer­ence between a Loan­able Funds view of the mon­e­tary sys­tem and that of Endoge­nous Money:

The Loan­able Funds model thus omits the con­tri­bu­tion of the change in debt to the level of aggre­gate demand.

  1. Occam’s Razor passes Endoge­nous Money & fails Loan­able Funds

If banks make loans to non-banks—as is man­i­festly the case—and cre­ate money in doing so by cred­it­ing the deposit accounts of their borrowers—as even the staunch advo­cate of Loan­able Funds Paul Krug­man has conceded—then the Loan­able Funds model is too extreme a sim­pli­fi­ca­tion of the nature of cap­i­tal­ism. As Ein­stein put it in rela­tion to physics:

It can scarcely be denied that the supreme goal of all the­ory is to make the irre­ducible basic ele­ments as sim­ple and as few as pos­si­ble with­out hav­ing to sur­ren­der the ade­quate rep­re­sen­ta­tion of a sin­gle datum of expe­ri­ence. (Ein­stein 1934, p. 165, empha­sis added)

Omit­ting the capac­ity of banks to cre­ate money, and the impact this has on key macro­eco­nomic aggre­gates omits a vital “datum of expe­ri­ence” from macro­eco­nomic mod­els. The capac­ity of bank lend­ing to alter the level of aggre­gate demand means that banks, debt and money must be included in any ade­quate model of macro­eco­nom­ics.

In par­tic­u­lar, the acknowl­edge­ment of the macro­eco­nomic sig­nif­i­cance of Endoge­nous Money requires a dynamic rede­f­i­n­i­tion of aggre­gate demand to include the change in debt. Though this model excludes sec­ond-order effects such as demand for idle cash bal­ances (Rowe 2013), the generic for­mula relat­ing aggre­gate demand (AD) to income (Y) and the change of debt is:

This for­mula cor­rects a rule of thumb propo­si­tion that I have pre­vi­ously asserted, that aggre­gate demand is the sum of income plus the change in debt (Keen 2014; see also Krug­man 2013b). The cor­rect propo­si­tion is that, in a world in which the bank­ing sec­tor endoge­nously cre­ates new money by cre­at­ing new loans, aggre­gate demand in a given period is the sum of aggre­gate demand at the begin­ning of that period, plus the change in debt over the period mul­ti­plied by the veloc­ity of money.

If we con­sider a time period of one year so that and , and spec­i­fy­ing the aver­age veloc­ity of money over that year as v(1) and the change in debt as DD(1), we have

Equa­tions and enable us to oper­a­tional­ize Keynes’s dis­tinc­tion between ex-ante and ex-post, while prov­ing the con­sis­tency of this dynamic for­mula with the stan­dard macro­eco­nomic account­ing iden­tity that expen­di­ture equals income. In words, these equa­tions assert that ex-post expen­di­ture equals ex-ante expen­di­ture (and hence income), plus the veloc­ity of money mul­ti­plied by the ex-post change in debt.

Since the veloc­ity of money com­fort­ably exceeds unity (though it is highly vari­able and pro-cycli­cal), the numer­i­cal impact of the change in debt on aggre­gate demand is there­fore larger than I have claimed in research prior to devel­op­ing this for­mal proof (Keen 2014; see also Rowe 2013).

  1. Sim­u­lat­ing Loan­able Funds and Endoge­nous Money

A sim­u­la­tion of the two mod­els con­firms the impor­tance of includ­ing the change in debt in aggre­gate demand. The sim­ple mod­els used here are iden­ti­cal except for the struc­ture of lend­ing, so that the dif­fer­ences in their behav­ior reflects sim­ply that issue. The mod­els use sim­ple vari­able time para­me­ters to relate the var­i­ous mon­e­tary flows to each other and the mon­e­tary stocks, so that the results do not depend on any behav­ioral assump­tions (see the Appen­dix for the model equa­tions and default para­me­ter val­ues). The val­ues of two of these parameters—the lend­ing and repay­ment rates—are var­ied over the sim­u­la­tions shown in Fig­ure 5 and Fig­ure 6.

Fig­ure 5: Loan­able Funds sim­u­la­tion in Min­sky

Fig­ure 6: Endoge­nous Money sim­u­la­tion in Min­sky

Vari­a­tions in the lend­ing and repay­ment rates have a minor effect on income in the Loan­able Funds model (see Fig­ure 7) because they impact upon the veloc­ity of cir­cu­la­tion of money (see Fig­ure 8). How­ever the level does not rise (or fall) sig­nif­i­cantly, and there is no trend, since vari­a­tions in the level of debt have no impact upon the money sup­ply, which remains con­stant (see Fig­ure 9).

Fig­ure 7: GDP as a func­tion of Lend­ing & Repay­ment rates in Loan­able Funds

Fig­ure 8: Money veloc­ity as a func­tion of Lend­ing & Repay­ment rates in Loan­able Funds

Fig­ure 9: Money and Debt as func­tions of Lend­ing & Repay­ment rates in Loan­able Funds

In con­trast, vari­a­tions in the lend­ing and repay­ment rates have a dra­matic impact upon GDP in the Endoge­nous Money model (see Fig­ure 10), because as well as hav­ing an impact upon the veloc­ity of money (see Fig­ure 11) they alter the rate of cre­ation and destruc­tion of money (see Fig­ure 12).

Fig­ure 10: GDP as a func­tion of Lend­ing & Repay­ment rates in Endoge­nous Money

Fig­ure 11: Money veloc­ity as a func­tion of Lend­ing & Repay­ment rates in Endoge­nous Money

Fig­ure 12: Money and Debt as func­tions of Lend­ing & Repay­ment rates in Endoge­nous Money

  1. Modeling financial instability

The pre­ced­ing proof pro­vides a the­o­ret­i­cal jus­ti­fi­ca­tion for the key role given to the level and change in aggre­gate pri­vate debt in Minsky’s Finan­cial Insta­bil­ity Hypoth­e­sis. Empir­i­cal research by Fama and French pro­vided fur­ther sup­port, by con­clud­ing that the cor­re­la­tions they found (includ­ing a 0.79 cor­re­la­tion between aggre­gate cor­po­rate invest­ment and change in long term cor­po­rate debt) “con­firm the impres­sion that debt plays a key role in accom­mo­dat­ing year-by-year vari­a­tion in invest­ment” (Fama and French 1999, p. 1954).

Min­sky pro­vided a suc­cinct sum­mary of his Finan­cial Insta­bil­ity Hypoth­e­sis, which empha­sized the cen­tral of pri­vate debt to his analy­sis (Min­sky 1978; reprinted in Min­sky 1982):

The nat­ural start­ing place for ana­lyz­ing the rela­tion between debt and income is to take an econ­omy with a cycli­cal past that is now doing well. The inher­ited debt reflects the his­tory of the econ­omy, which includes a period in the not too dis­tant past in which the econ­omy did not do well. Accept­able lia­bil­ity struc­tures are based upon some mar­gin of safety so that expected cash flows, even in peri­ods when the econ­omy is not doing well, will cover con­trac­tual debt pay­ments. As the period over which the econ­omy does well length­ens, two things become evi­dent in board rooms. Exist­ing debts are eas­ily val­i­dated and units that were heav­ily in debt pros­pered; it paid to lever. After the event it becomes appar­ent that the mar­gins of safety built into debt struc­tures were too great. As a result, over a period in which the econ­omy does well, views about accept­able debt struc­ture change. In the deal­mak­ing that goes on between banks, invest­ment bankers, and busi­ness­men, the accept­able amount of debt to use in financ­ing var­i­ous types of activ­ity and posi­tions increases. This increase in the weight of debt financ­ing raises the mar­ket price of cap­i­tal assets and increases invest­ment. As this con­tin­ues the econ­omy is trans­formed into a boom econ­omy.

Sta­ble growth is incon­sis­tent with the man­ner in which invest­ment is deter­mined in an econ­omy in which debt-financed own­er­ship of cap­i­tal assets exists, and the extent to which such debt financ­ing can be car­ried is mar­ket deter­mined. It fol­lows that the fun­da­men­tal insta­bil­ity of a cap­i­tal­ist econ­omy is upward. The ten­dency to trans­form doing well into a spec­u­la­tive invest­ment boom is the basic insta­bil­ity in a cap­i­tal­ist econ­omy. (Min­sky 1982, pp. 66–67)

I mod­eled this process by extend­ing Goodwin’s cycli­cal growth model—in which profit-rate-moti­vated invest­ment and employ­ment-rate-moti­vated wage demands gen­er­ated a closed limit cycle in employ­ment and income dis­tri­b­u­tion (Good­win 1967)—to include debt-financed invest­ment. Goodwin’s model reduced to two cou­pled dif­fer­en­tial equa­tions in the employ­ment rate (?) and wages share of out­put (?), where is a Phillips-curve rela­tion and is an invest­ment func­tion depend­ing on the rate of profit :

I replaced Goodwin’s “starkly schema­tized” (Good­win 1967, p. 54) assump­tion that invest­ment equalled profit at all times with an invest­ment func­tion in which invest­ment exceeded profit at high rates of profit, and was below profit at low rates. An equa­tion to rep­re­sent debt-financed invest­ment was added—Equation —and profit was rede­fined as earn­ings net of inter­est pay­ments :

This trans­formed Goodwin’s model into a three-state model of Minsky’s hypoth­e­sis, with the extra equa­tion being the dynam­ics of the pri­vate debt to out­put ratio (see Keen 2013, pp. 236–38 for the deriva­tion):

In (Keen 1995; Keen 2000) I used non­lin­ear func­tions for both invest­ment deter­mi­na­tion and wage set­ting; here I use lin­ear func­tions to empha­size that both the cycli­cal behav­ior of Goodwin’s model and the debt-induced break­down in the Min­sky model are endemic, rather than being prod­ucts of the assumed func­tional forms. In the sim­u­la­tions shown in Fig­ure 13 and Fig­ure 14, the invest­ment and wage change func­tions are:

Fig­ure 13 shows the fixed cycle in Goodwin’s basic model.

Fig­ure 13: Goodwin’s model with lin­ear behav­ioral func­tions sim­u­lated in Min­sky

Fig­ure 14 shows a typ­i­cal run of the Min­sky model, which has three key char­ac­ter­is­tics:

  • The ini­tial behav­ior of the model involves a reduc­tion in the volatil­ity of employ­ment and output—effectively a “Great Mod­er­a­tion”;
  • Work­ers’ share of out­put has a sec­u­lar ten­dency to fall; and
  • The ini­tial reduc­tion in employ­ment and out­put volatil­ity gives way to increas­ing volatil­ity as the debt to out­put level rises (with the ulti­mate out­come of a debt-induced col­lapse in out­put and employ­ment).

Fig­ure 14: Minsky’s FIH with lin­ear behav­ioral func­tions sim­u­lated in Min­sky

The fact that this sim­ple model gen­er­ated out­comes that, in a very styl­ized way, mir­ror the empir­i­cal record of the recent eco­nomic past, empha­sizes the impor­tance of devel­op­ing an approach to macro­eco­nom­ics in which banks and pri­vate debt play inte­gral roles. The empir­i­cal data, inter­preted in the light of the the­o­ret­i­cal argu­ments given here, fur­ther empha­sizes the impor­tance of pay­ing close pol­icy atten­tion to the hith­erto ignored phe­nom­e­non of the growth of pri­vate debt.

  1. Empirical Data

For­tu­nately, though main­stream eco­nomic the­ory has ignored the role of pri­vate debt, sta­tis­ti­cal agen­cies have col­lected the data. Fig­ure 15 is an imputed series com­bin­ing actual Fed­eral Reserve quar­terly data on house­hold plus non-finan­cial cor­po­rate debt since 1952 (and yearly data from 1945 till 1952) with US Cen­sus data from 1916–1970, and par­tial Cen­sus data on bank loans from 1834 to 1970 (Cen­sus 1949; Cen­sus 1975).

Fig­ure 15: US pri­vate debt since 1834

The causal role of the change in debt in aggre­gate demand iden­ti­fied in this paper implies that there should be a strong empir­i­cal rela­tion­ship between change in debt and macro­eco­nomic data such as the unem­ploy­ment rate—in con­trast to the Loan­able-Funds-based pre­sump­tion that “Absent implau­si­bly large dif­fer­ences in mar­ginal spend­ing propen­si­ties among the groups … pure redis­tri­b­u­tions should have no sig­nif­i­cant macro-eco­nomic effects…” (Bernanke 2000, p. 24). This Loan­able Funds pre­sump­tion is strongly rejected by the data. As Fig­ure 16 shows, the cor­re­la­tion of the change in debt times veloc­ity (divided by GDP) with the level of unem­ploy­ment since 1990 is –0.92.

Fig­ure 16: Change in debt times veloc­ity and US Unem­ploy­ment (Cor­re­la­tion –0.92)

The first dif­fer­ence of also implies a strong rela­tion­ship between the change in the change in debt over two time peri­ods and change in unem­ploy­ment over that period. Set­ting , the change in aggre­gate demand between peri­ods t2–t1 and t1–t0 (nor­mal­ized by divid­ing by ) is:

Set­ting , the cor­re­la­tion between equa­tion , which we term the Credit Accel­er­a­tor (see also Biggs and Mayer 2010; Biggs, et al. 2010), and the annual per­cent­age change in the unem­ploy­ment rate over the period from 1975 till today is –0.78 (see Fig­ure 17).

Fig­ure 17: Credit accel­er­a­tion and change in unem­ploy­ment (Cor­re­la­tion –0.78)

  1. Conclusion

Given that bank lend­ing cre­ates money and repay­ment of debt destroys it, the change in debt plays an inte­gral role in macro­eco­nom­ics by dynam­i­cally vary­ing the level of aggre­gate demand. The omis­sion of this fac­tor from main­stream eco­nomic mod­els is the rea­son that these mod­els failed to warn of the dan­gers of the dra­matic buildup in pri­vate debt since WWII—and espe­cially since 1993, when the debt-financed recov­ery from the 1990s reces­sion took the aggre­gate pri­vate debt level past the peak caused by defla­tion in the 1930s (see Fig­ure 15). It is also the rea­son why they failed to antic­i­pate the cri­sis that began in 2007, and instead pre­dicted that, as the OECD put it in June 2007, “the cur­rent eco­nomic sit­u­a­tion is in many ways bet­ter than what we have expe­ri­enced in years… Our cen­tral fore­cast remains indeed quite benign” (OECD 2007). Pol­icy mak­ers rely­ing upon main­stream econ­o­mists as experts on the func­tion­ing of the econ­omy thus not only received no warn­ing about the worst eco­nomic cri­sis since the Great Depres­sion, but were falsely led to expect benign rather than malig­nant eco­nomic con­di­tions.

The erro­neous neglect of the dynam­ics of pri­vate debt by the eco­nom­ics pro­fes­sion has there­fore resulted in enor­mous social and eco­nomic harm to soci­ety. This is the oppo­site of the intended goal of eco­nomic the­ory and pol­icy. If eco­nomic the­ory and pol­icy are to ful­fil their intended role, it is imper­a­tive that a reformed macro­eco­nom­ics be devel­oped in which banks, money and the dynam­ics of debt play inte­gral roles.

  1. Appendix

    1. Loanable Funds model

Differential equations for money and debt

Other differential equations

  1. Endogenous Money model

Differential equations for money and debt

Other differential equations

Common Definitions

Common Parameters to Loanable Funds and Endogenous Money models

  1. Goodwin model

  1. Minsky model (new and modified equations only)

  1. Common parameters to Goodwin & Minsky models

  1. Minsky

Min­sky is an addi­tion to the fam­ily of sys­tem dynam­ics pro­grams that began with Jay Forrester’s pio­neer­ing work on devel­op­ing a visual metaphor for con­struct­ing and sim­u­lat­ing dynamic mod­els of com­plex social and eco­nomic processes (For­rester 1968). Forrester’s metaphor was the flow­chart (see Fig­ure 18): a draw­ing of the rela­tion­ships in a sys­tem became the frame­work for devel­op­ing a math­e­mat­i­cal model of that sys­tem:

The pro­posed model struc­ture and method of solu­tion retain a one-to-one cor­re­spon­dence between the pre­sumed form of the real eco­nomic world and the quan­ti­ties, coef­fi­cients, vari­ables, and deci­sion cri­te­ria of the model. For­mu­la­tion in terms of a “flow dia­gram” is pos­si­ble so that a pic­to­r­ial rep­re­sen­ta­tion of the rela­tion­ships within the sys­tem is avail­able at all times. (For­rester 2003p. 344 )

Fig­ure 18: The first sys­tem dynam­ics dia­gram from For­rester 2003 (1956)

There are now at least a dozen pro­grams imple­ment­ing this mod­el­ing phi­los­o­phy, rang­ing from the free Open Source pro­gram Xcos to the $20,000-a-copy com­mer­cial pro­gram Simulink. This par­a­digm is now per­va­sive in engi­neer­ing, but it failed to take root in eco­nom­ics, despite the fact that Forrester’s con­cept was twice antic­i­pated in economics—firstly by Irv­ing Fisher in 1891 with a hydraulic model for cal­cu­lat­ing equi­lib­rium val­ues in a Wal­rasian model (Brainard and Scarf), and then by the engi­neer-turned econ­o­mist Bill Phillips with gen­uinely dynamic ana­log com­puter sys­tems (Hayes 2011; Lee­son 1994a; Lee­son 1994b; Lee­son 1995; Lee­son 2000; Phillips 1950; Phillips 1954; Phillips 1957) some years before For­rester. How­ever, there was no devel­op­ment in eco­nom­ics com­pa­ra­ble to Forrester’s inno­va­tion (in con­junc­tion with the com­puter pro­gram­mers Phyl­lis Fox and Alexan­der Pugh–see Lane 2007) of a dig­i­tal com­puter program—DYNAMO—to pro­vide a gen­eral pur­pose foun­da­tion for build­ing dynamic mod­els of com­plex sys­tems.

Fig­ure 19: Fisher’s 1891 hydraulic machine for cal­cu­lat­ing Wal­rasian equi­lib­rium prices, from Brainard and Scarf, p. 69

Fig­ure 20: Phillips’s schematic dia­gram of a dynamic mul­ti­plier-accel­er­a­tor model, from Phillips 1954, p. 306

The core par­a­digm in sys­tem dynam­ics pro­grams is the con­struc­tion of math­e­mat­i­cal equa­tions via flow­charts iden­ti­cal in spirit to that devel­oped by Phillips (see Fig­ure 20). For exam­ple, Fig­ure 21 is the sys­tem dynam­ics equiv­a­lent of the dif­fer­en­tial equa­tion for expo­nen­tial pop­u­la­tion growth .

Fig­ure 21: A sim­ple alge­braic equa­tion in a sys­tem dynam­ics pro­gram (Min­sky)

Sim­ple expres­sions like this are just as eas­ily ren­dered in equa­tions or stan­dard text-ori­ented com­puter pro­grams, but the sys­tem dynam­ics approach makes it eas­ier to com­pre­hend much more com­plex models—hence its dom­i­nance in the engi­neer­ing field today.

Min­sky pro­vides this clas­sic sys­tem dynam­ics approach, and also adds a new method of con­struct­ing dif­fer­en­tial equa­tions to the sys­tem dynam­ics toolkit that is supe­rior for mod­el­ling finan­cial flows: the God­ley Table. Based on the account­ing con­cept of dou­ble-entry book­keep­ing, each col­umn rep­re­sents the dynamic equa­tion of a given finan­cial account, while each row rep­re­sents trans­ac­tions between accounts. This is a more nat­ural way to por­tray finan­cial trans­ac­tions which also helps enforce the fun­da­men­tal rules of accounting—that Assets equal Lia­bil­i­ties plus Equity.

Min­sky ensures this in three ways. Firstly, all row oper­a­tions in a God­ley Table must sum to zero—otherwise an error is flagged. Sec­ondly, the source of any trans­ac­tion is shown as a pos­i­tive while the des­ti­na­tion (or “sink” in sys­tem dynam­ics par­lance) is shown as a neg­a­tive. Thirdly, Assets are shown as pos­i­tive while Lia­bil­i­ties and Equity are shown as neg­a­tive. Fig­ure 22 illus­trates these three conventions—including show­ing what hap­pens when they are breached.

Fig­ure 22: A sam­ple God­ley Table


Bernanke, Ben S.

    2000    Essays on the Great Depres­sion. Prince­ton: Prince­ton Uni­ver­sity Press.

Beze­mer, Dirk J

    2009    “No One Saw This Com­ing”: Under­stand­ing Finan­cial Cri­sis Through Account­ing Mod­els. Gronin­gen, The Nether­lands: Fac­ulty of Eco­nom­ics Uni­ver­sity of Gronin­gen.

Biggs, Michael, and Thomas Mayer

    2010    The Out­put Gap Conun­drum. Intereconomics/Review of Euro­pean Eco­nomic Pol­icy 45(1):11–16.

Biggs, Michael, Thomas Mayer, and Andreas Pick

    2010    Credit and Eco­nomic Recov­ery: Demys­ti­fy­ing Phoenix Mir­a­cles. SSRN eLi­brary.

Blan­chard, Olivier

    2009    The State of Macro. Annual Review of Eco­nom­ics 1(1):209–228.

Blan­chard, Olivier, Gio­vanni Dell’Ariccia, and Paolo Mauro

    2010    Rethink­ing Macro­eco­nomic Pol­icy. Jour­nal of Money, Credit, and Bank­ing 42:199–215.

Brainard, William C., and Her­bert E. Scarf

        How to Com­pute Equi­lib­rium Prices in 1891. Amer­i­can Jour­nal of Eco­nom­ics and Soci­ol­ogy 64:57–83.

Car­ney, John

    2012    What Really Con­strains Bank Lend­ing. In Net­Net. J. Car­ney, ed. New York: CNBC.

Car­pen­ter, Seth B., and Selva Demi­ralp

    2010    Money, Reserves, and the Trans­mis­sion of Mon­e­tary Pol­icy: Does the Money Mul­ti­plier Exist? In Finance and Eco­nom­ics Dis­cus­sion Series. Wash­ing­ton: Fed­eral Reserve Board.

Cen­sus, Bureau of

    1949    His­tor­i­cal Sta­tis­tics of the United States 1789–1945. B.o.t. Cen­sus, ed. Wash­ing­ton: United States Gov­ern­ment.

    1975    His­tor­i­cal Sta­tis­tics of the United States Colo­nial Times to 1970. B.o.t. Cen­sus, ed. Wash­ing­ton: United States Gov­ern­ment.


    2012    Mon­e­tary and Finan­cial Devel­op­ments. In Monthly Bul­letin May 2012. Brus­sels: Euro­pean Cen­tral Bank.

Eggerts­son, Gauti B., and Paul Krug­man

    2012a    Debt, Delever­ag­ing, and the Liq­uid­ity Trap: A Fisher-Min­sky-Koo approach. Quar­terly Jour­nal of Eco­nom­ics 127:1469–1513.

    2012b    Sup­ple­men­tary mate­r­ial to Debt, Delever­ag­ing and the Liq­uid­ity Trap. Quar­terly Jour­nal of Eco­nom­ics 127:Appendix.

Ein­stein, Albert

    1934    On the Method of The­o­ret­i­cal Physics. Phi­los­o­phy of Sci­ence 1:163–169.

Fama, Eugene F., and Ken­neth R. French

    1999    The Cor­po­rate Cost of Cap­i­tal and the Return on Cor­po­rate Invest­ment. Jour­nal of Finance 54(6):1939–1967.

Fisher, Irv­ing

    1933    The Debt-Defla­tion The­ory of Great Depres­sions. Econo­met­rica 1(4):337–357.

For­rester, Jay W.

    1968    Indus­trial Dynam­ics-After the First Decade. Man­age­ment Sci­ence 14(7):398–415.

    2003    Dynamic mod­els of eco­nomic sys­tems and indus­trial orga­ni­za­tions: Note to the Fac­ulty research sem­i­nar. From Jay W. For­rester. Novem­ber 5, 1956. Sys­tem Dynam­ics Review 19(4):329–45.

Fried­man, Mil­ton

    1948    A Mon­e­tary and Fis­cal Frame­work for Eco­nomic Sta­bil­ity. The Amer­i­can Eco­nomic Review 38(3):245–264.

    1959    The Demand for Money: Some The­o­ret­i­cal and Empir­i­cal Results. The Amer­i­can Eco­nomic Review 49(2):525–527.

    1969    The Opti­mum Quan­tity of Money. In The Opti­mum Quan­tity of Money and Other Essays. Pp. 1–50. Chicago: MacMil­lan.

Fried­man, Mil­ton, and Anna Jacob­son Schwartz

    1963    A mon­e­tary his­tory of the United States 1867–1960. Prince­ton: Prince­ton Uni­ver­sity Press.

Full­wiler, Scott

    2012    Krugman’s Flash­ing Neon Sign. In New Eco­nomic Per­spec­tives. S. Kel­ton, ed. Kansas City: Stephanie Kel­ton.

Good­win, Richard M.

    1967    A growth cycle. In Social­ism, Cap­i­tal­ism and Eco­nomic Growth. C.H. Fein­stein, ed. Pp. 54–58. Cam­bridge: Cam­bridge Uni­ver­sity Press.

Graziani, Augusto

    1989    The The­ory of the Mon­e­tary Cir­cuit. Thames Papers in Polit­i­cal Econ­omy Spring:1–26.

Hayes, Brian

    2011    Eco­nom­ics, Con­trol The­ory, and the Phillips Machine. ASSRU — Algo­rith­mic Social Sci­ence Research Unit.

Holmes, Alan R.

    1969    Oper­a­tional Con­straints on the Sta­bi­liza­tion of Money Sup­ply Growth. In Con­trol­ling Mon­e­tary Aggre­gates. F.E. Mor­ris, ed. Pp. 65–77. Nan­tucket Island: The Fed­eral Reserve Bank of Boston.

Keen, Steve

    1995    Finance and Eco­nomic Break­down: Mod­el­ing Minsky’s ‘Finan­cial Insta­bil­ity Hypoth­e­sis.’. Jour­nal of Post Key­ne­sian Eco­nom­ics 17(4):607–635.

    2000    The Non­lin­ear Eco­nom­ics of Debt Defla­tion. In Com­merce, com­plex­ity, and evo­lu­tion: Top­ics in eco­nom­ics, finance, mar­ket­ing, and man­age­ment: Pro­ceed­ings of the Twelfth Inter­na­tional Sym­po­sium in Eco­nomic The­ory and Econo­met­rics. W.A. Bar­nett, C. Chiarella, S. Keen, R. Marks, and H. Schn­abl, eds. Pp. 83–110. New York: Cam­bridge Uni­ver­sity Press.

    2013    Pre­dict­ing the ‘Global Finan­cial Cri­sis’: Post-Key­ne­sian Macro­eco­nom­ics. Eco­nomic Record 89(285):228–254.

    2014    Sec­u­lar stag­na­tion and endoge­nous money. Real World Eco­nom­ics Review 66:2–11.

Keis­ter, T., and J. McAn­drews

    2009    Why Are Banks Hold­ing So Many Excess Reserves? New York: Fed­eral Reserve Bank of New York.

Krug­man, Paul

    2012a    Bank­ing Mys­ti­cism. In The Con­science of a Lib­eral. New York: New York Times.

    2012b    Bank­ing Mys­ti­cism, Con­tin­ued. In The Con­science of a Lib­eral. New York: New York Times.

    2012c    End this Depres­sion Now! New York: W.W. Nor­ton.

    2012d    Min­sky and Method­ol­ogy (Wonk­ish). In The Con­science of a Lib­eral. New York: New York Times.

    2012e    Oh My, Steve Keen Edi­tion. In The Con­science of a Lib­eral. New York: New York Times.

    2012f    Things I Should Not Be Wast­ing Time On. In The Con­science of a Lib­eral. New York: New York Times.

    2013a    Com­mer­cial Banks As Cre­ators of “Money”. In The Con­science of a Lib­eral.

    2013b    Sec­u­lar Stag­na­tion Arith­metic. In The Con­science of a Lib­eral. P. Krug­man, ed. New York: New York Times.

Kyd­land, Finn E., and Edward C. Prescott

    1990    Busi­ness Cycles: Real Facts and a Mon­e­tary Myth. Fed­eral Reserve Bank of Min­neapo­lis Quar­terly Review 14(2):3–18.

Lane, David C.

    2007    The power of the bond between cause and effect: Jay Wright For­rester and the field of sys­tem dynam­ics. Sys­tem Dynam­ics Review 23:95–118.

Lee­son, Robert

    1994a    Some Mis­un­der­stand­ings Con­cern­ing the Con­tri­bu­tions Made by A. W. H. Phillips and R. G. Lipsey to the Infla­tion-Unem­ploy­ment Lit­er­a­ture. His­tory of Eco­nom­ics Review (22):70–82.

    1994b    A. W. H. Phillips M.B.E. (Mil­i­tary Divi­sion). Eco­nomic Jour­nal 104(424):605–618.

    1995    A. W. H. Phillips: His Machine and His Curve. New Zealand Eco­nomic Papers 29(2):231–243.

    2000    A. W. H. Phillips: An Extra­or­di­nary Life. In A. W. H. Phillips: Col­lected works in con­tem­po­rary per­spec­tive. R. Lee­son, ed. Pp. 3–17. Cam­bridge: Cam­bridge Uni­ver­sity Press.

Min­sky, Hyman

    1978    The Finan­cial Insta­bil­ity Hypoth­e­sis: A Restate­ment. Thames Papers in Polit­i­cal Econ­omy Autumn 1978.

Min­sky, Hyman P.

    1980    Cap­i­tal­ist Finan­cial Processes and the Insta­bil­ity of Cap­i­tal­ism. Jour­nal of Eco­nomic Issues 14(2):505–523.

    1982    Can “it” hap­pen again? : essays on insta­bil­ity and finance. Armonk, N.Y.: M.E. Sharpe.

Moore, Basil J.

    1979    The Endoge­nous Money Stock. Jour­nal of Post Key­ne­sian Eco­nom­ics 2(1):49–70.

    1983    Unpack­ing the Post Key­ne­sian Black Box: Bank Lend­ing and the Money Sup­ply. Jour­nal of Post Key­ne­sian Eco­nom­ics 5(4):537–556.


    2007    Achiev­ing Fur­ther Rebal­anc­ing. In OECD Eco­nomic Out­look. OECD, ed: OECD Pub­lish­ing.

Pal­ley, Thomas

    2013    Hor­i­zon­tal­ists, ver­ti­cal­ists, and struc­tural­ists: the the­ory of endoge­nous money reassessed. Review of Key­ne­sian Eco­nom­ics 1(4):406–424.

Pal­ley, Thomas I.

    1991    The Endoge­nous Money Sup­ply: Con­sen­sus and Dis­agree­ment. Jour­nal of Post Key­ne­sian Eco­nom­ics 13(3):397–403.

    2002    Endoge­nous Money: What It Is and Why It Mat­ters. Metroe­co­nom­ica 53(2):152–180.

Phillips, A. W.

    1950    Mechan­i­cal Mod­els in Eco­nomic Dynam­ics. Eco­nom­ica 17(67):283–305.

    1954    Sta­bil­i­sa­tion Pol­icy in a Closed Econ­omy. The Eco­nomic Jour­nal 64(254):290–323.

    1957    Sta­bil­i­sa­tion Pol­icy and the Time-Forms of Lagged Responses. The Eco­nomic Jour­nal 67(266):265–277.

Roche, Cullen

    2013    Keep Banks Out of Macro? In Prag­matic Cap­i­tal­ism. C. Roche, ed. New York: Orcam Finan­cial Group, LLC.

Rowe, Nick

    2013    What Steve Keen is maybe try­ing to say. In Worth­while Cana­dian Ini­tia­tive. N. Rowe, ed. Canada.

Smets, Frank, and Rafael Wouters

    2007    Shocks and Fric­tions in US Busi­ness Cycles: A Bayesian DSGE Approach. Amer­i­can Eco­nomic Review 97(3):586–606.

Solow, Robert M.

    2003    Dumb and Dumber in Macro­eco­nom­ics. In Festschrift for Joe Stiglitz. Colum­bia Uni­ver­sity.

    2008    The State of Macro­eco­nom­ics. The Jour­nal of Eco­nomic Per­spec­tives 22(1):243–246.

Sum­ner, Scott

    2013    Keep banks out of macro. In The Money Illu­sion. Bent­ley: Scott Sum­ner.

Tobin, J.

    1963    Com­mer­cial banks as cre­ators of “money”. In Bank­ing and Mon­e­tary Stud­ies. D. Car­son, ed. Pp. 408–419. Wash­ing­ton: Comp­trol­ler of the Cur­rency.

Wood­ford, Michael

    2009    Con­ver­gence in Macro­eco­nom­ics: Ele­ments of the New Syn­the­sis. Amer­i­can Eco­nomic Jour­nal: Macro­eco­nom­ics 1(1):267–79.



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.
  • Steve Hum­mel

    Excel­lent conclusions…as far as they go. I sug­gest you study C. H. Dou­glas, a con­tem­po­rary of Keynes, who came to the same and even deeper finan­cial and mon­e­tary insta­bil­ity con­clu­sions as your­self. Keynes’ solu­tions were actu­ally just pal­lia­tives which enabled the finan­cial elites to bas­tardize them fur­ther with the help of neo-clas­si­cal econ­o­mists and their obscur­ing notions of money being “a veil over barter”.

  • Pingback: Modeling Financial Instability | Heterodox econ...()

  • koonyeow

    I sup­pose the next log­i­cal step is to observe if gov­ern­ments accept the Finan­cial Insta­bil­ity Hypoth­e­sis.

  • Rob Rawl­ings

    What is described here is log­i­cally con­sis­tent model that does show that if banks cre­ated money out of thin air then the change in demand would indeed be a func­tion of the change in debt and the veloc­ity of money.

    I have some doubt how well this maps on to the real econ­omy.

    Banks can cre­ate new loans with­out need­ing to first have fund avail­able to lend. How­ever they do need to have funds avail­able to set­tle trans­fers to other banks if and when any of these new loans are spent. Govt base money is used for this. 

    Assum­ing no reserve require­ment banks will keep the opti­mum level of base money in reserves to set­tle with other banks (with inter­bank lend­ing and the dis­count win­dow allow­ing some flex­i­bil­ity). Given a fixed base and assum­ing the amount of lend­ing in the bank­ing sys­tem is fully opti­mized then the money sup­ply will be a func­tion of bank reserves and money veloc­ity since there is pos­i­tive cor­re­la­tion between veloc­ity and reserve require­ments. If the base doesn’t change then the main dri­ver of an increased money sup­ply will be change in veloc­ity. If the mon­e­tary author­i­ties wish to increase the money sup­ply at a time when veloc­ity is con­stant they will do so by increas­ing the base (by low­er­ing the tar­get rate for inter­bank lend­ing).

    So in the real econ­omy a change in AD will be dri­ven either by a change in veloc­ity or a change in base money. Other things equal lend­ing will be deter­mined by these 2 things.

  • Rob Rawl­ings

    the money sup­ply will be a func­tion of bank reserves and money veloc­ity”

    should read

    the money sup­ply will be a func­tion of base money and money veloc­ity ”

  • Steve Hum­mel

    I con­sider veloc­ity to actu­ally be largely if not com­pletely the result of the amount of con­tin­u­ous lend­ing as there is close to a 100% cor­re­la­tion between the nom­i­nal amount of lend­ing, the fall of the rate of veloc­ity and a reces­sion or depres­sion. Also, even if one assumes mul­ti­ple re-cir­cu­la­tions of money within the repay­ment of each finan­cial cycle veloc­ity still abstracts out the cost account­ing real­i­ties that are endemic to/utterly embed­ded within the economy/productive process and hence, all fac­tors con­sid­ered, there is always a higher rate of flow of MINIMAL total prices than there is a rate of flow of indi­vid­ual pur­chas­ing power (wages minus taxes). Veloc­ity is a poorly con­structed the­ory and even if expanded as per above.…consistently if not per­pet­u­ally fails to equate total indi­vid­ual pur­chas­ing power and total prices. 

    Add to the above that the cap­i­tal costs of increas­ing high tech­nol­ogy are ris­ing while it is simul­ta­ne­ously reduc­ing the ratio­nal need for human effort and even human input, (wages).…and you’ve got an eco­nomic conun­drum which is only reme­di­a­ble by adding to the cur­rent con­sumer finan­cial par­a­digms of work for pay and loan ONLY, the con­cept of mon­e­tary grace the free gift, which increases indi­vid­ual pur­chas­ing power with­out incur­ring an addi­tional cost.

  • Paul Andrews

    Pri­vate debt opti­mizes con­trol of pro­duc­tive resources — the owner is rarely the best con­troller and pri­vate debt is the main mech­a­nism by which con­trol is ceded but own­er­ship retained. Lack of it is fatal. Too much of it (via dis­tor­tion) deop­ti­mizes.

    The macro­eco­nomic effects are obvi­ously highly sig­nif­i­cant. Research sug­gest­ing oth­er­wise is funded by those who ben­e­fit from the dis­tor­tions.

  • Pingback: Anonymous()

  • Bhaskara II

    I’ll bet some of you would like these graphs! Espe­cially you pro­fes­sor Keen, since you men­tioned, “The fact that this sim­ple model gen­er­ated out­comes that, in a very styl­ized way, mir­ror the empir­i­cal record of the recent eco­nomic past,”

    Here are empir­i­cal plots for what is in Fig­ure 14 for the coun­try of Aus­tralia. The last year or two are prob­a­bly pro­jec­tions.

    The left hand plot is Wages Share of GDP in Per­cent (total wages/GDP) and employ­ment pop­u­la­tion ratio not in per­cent but a ratio. A com­mon pat­tern is a gen­eral counter clock­wise motion often loop­ing counter clock­wise while mov­ing to dif­fer­ent places. The Good­win model’s loop direc­tion matches the result. Keen’s sim­u­la­tion goes the oppo­site direc­tion because the vari­ables are on swapped axises (think about it or sketch it). The decades each have a color and the date of every 5 year points are labeled.

    The decade col­ors are in the order of pink, light blue, dark brown, light orange, orange, and red.

    With out being uptight about it match­ing a sim­ple model here are my obser­va­tions of the data.

    There is CCW loop­ing motion for peri­ods of about 6 years cen­tered near ( pos­si­bly 1963), 1975, 1985, 1990, 2010. Pos­si­bly other peri­ods also.

    Counter clock­wise loop­ing indi­cates the X axis vari­able lead­ing the Y axis vari­able. Clock­wise would mean the Y axis vari­able lead­ing the X axis vari­able. In this case the employ­ment ratio often leads wages share while in a loop. 

    The loops are often near a 6 year period for many coun­tries. If I remem­ber right Jim Rogers claims the eco­nomic cycle dura­tion is 4 to 7 or 8 years. 

    The right hand plot has both vari­ables in per­cent on y axis and the date on the x axis. Pleas ignore the wrong labels for the right hand plot.

    I did these for about 20 coun­tries. I won­der if it would be of tan­gi­ble value to any one?

  • Bhaskara II

    Typo cor­rec­tion:

    Should have said, ” Please ignore the wrong labels for the right hand plot.”

    Other wise pretty cool data and analy­sis.

  • Bhaskara II

    I get a bet­ter clearer and in my opin­ion more accu­rate result than many of the other empirical-“Goodwin cycle” papers avail­able on line, for var­i­ous rea­sons.

  • math­ieu dufresne

    Nice paper, I think you’re head­ing the right way. How­ever, I’m not fully com­fort­able with equa­tion 0.15, altough I agree with the under­ly­ing intu­ition. For this equa­tion to be cor­rect, you need to make two very strong assump­tions.

    First, you need to assume that veloc­ity is con­stant across this period and the pre­vi­ous one, as the equa­tion doesn’t take into account the impact of a change in veloc­ity on the already exist­ing stock of money. This could be eas­ily taken into account but the next one is more tricky. 

    You also have to assume that all debt is incurred at the begin­ing of the period. For exam­ple, if a debt is taken near the end of a period, it’s impact on AD will be smaller than delta D times veloc­ity. (I assume here that by veloc­ity you mean a turnover rate rather than a free vari­able that is there to bal­ance any given equa­tion)

  • Rob Rawl­ings

    Change in AD = (change in M) * (change in V) is true by tau­tol­ogy.

    I can think of at least 3 rea­sons why M may change

    1. CB bal­ance sheet changes (ie: the base changes)
    2. The govt runs a deficit or sur­plus
    3. Banks expand lend­ing into related to 1. and 2.

    If 1 and 2 are all are assumed to stay con­stant then Steve’s for­mula

    Change in AD =(change in lend­ing) * (change in V) would be true.


    In the real econ­omy changes in 1. and 2. will be fre­quently and when they do change then this will (espe­cially out­side of ZLB) also drive changes in 3. 

    So: While changes in banks lend­ing poli­cies will have some con­tri­bu­tion to increases in AD 

    to say:

    aggre­gate demand in a given period is the sum of aggre­gate demand at the begin­ning of that period, plus the change in debt over the period mul­ti­plied by the veloc­ity of money”

    seems an over­sim­pli­fi­ca­tion.

  • Rob Rawl­ings


    (change in V) should just be V.

  • Bhaskara II

    Per­fect car­toon to go with loop­ing graph above for employ­ment rate and wages share:

    “We’re still not sure what hap­pened here, but I think we can all agree that we’re glad it’s over.”

  • Pingback: Modeling financial instability | Real-World Economics Review Blog()

  • ceviche

    I have a seman­tic con­cern with “bank lend­ing cre­ates money and repay­ment of debt destroys it” not because it is wrong but because it sends the wrong mes­sage to those who “don’t get it”, and con­fuse money cre­ation and destruc­tion with seignor­age.
    What i would pre­fer is
    “The finan­cial sec­tor is mod­elled as qual­i­ta­tively dif­fer­ent from the rest of the econ­omy and the sup­ply of money is treated as largely endoge­nous. So an increase in the money sup­ply is actu­ally often a mis­nomer as what is being said is really that there has been an expan­sion in credit rights or debt oblig­a­tions between par­tic­u­lar par­ties, usu­ally between a cen­tral bank and com­mer­cial banks or between the lat­ter and their cus­tomers. When a debt is repaid no money is destroyed, rather both an oblig­a­tion and a right are expunged. ”

  • Pingback: Debtonomics: Currency Crisis | Economic Undertow()

  • Pingback: Debtonomics: Currency Crisis | Economic Undertow()

  • Pingback: The credit impulse strengthens in March 2014 | The Macroeconomic Project()

  • Pingback: What Would We Like to See on the Curriculum? | Post-Crash Economics Society()

  • Pingback: To save humanity and the planet, we must redesign money - StuntFM 97.3()