Spe­cific projects

The Cen­ter for Eco­nomic Sta­bil­ity Incorporated

With the sup­port of blog mem­bers, I have formed the Cen­ter for Eco­nomic Sta­bil­ity Incor­po­rated. Our objec­tive is to develop CfESI into an empirically-oriented think-tank on eco­nom­ics that will develop real­is­tic analy­sis of cap­i­tal­ism, and pro­mote poli­cies based upon that analy­sis. The suc­cess of CfESI is depen­dent upon rais­ing suf­fi­cient fund­ing to enable staff to be hired who can take over the admin­is­tra­tive and web duties from me, and sup­ple­ment my research efforts.


Named in honor of Hyman Min­sky, this is a com­puter pro­gram that enables a com­plex mon­e­tary sys­tem to be mod­elled with rel­a­tive ease. The pro­gram imple­ments the tab­u­lar approach to mod­el­ling finan­cial flows devel­oped in (Keen 2008; Keen 2010Keen 2011), and com­bines this with the “flow­chart” par­a­digm devel­oped by engi­neers to model phys­i­cal processes, and imple­mented in numer­ous soft­ware pro­grams (Simulink, Vis­sim, Ven­sim, Ithink, Stella, etc.). It will be both a ped­a­gogic tool to make dynamic mon­e­tary mod­el­ling easy and attrac­tive to new stu­dents, and a pow­er­ful research tool that will enable the con­struc­tion of real­is­tic, mon­e­tary mod­els of capitalism.

Fig­ure 5

  • A first ver­sion of Min­sky is already under devel­op­ment, with fund­ing pro­vided by a grant from the Insti­tute for New Eco­nomic Think­ing. This ver­sion, to be com­pleted in mid-2012, will enable the mod­el­ling of the econ­omy as a mon­e­tary dynamic sin­gle com­mod­ity sys­tem. A pro­to­type will be released in early 2012. A Source­forge page is now oper­at­ing, and we will shortly be open­ing it up for col­lab­o­ra­tion by Open Source developers.
  • Ver­sion 2.0 will enable multi-commodity input-output dynam­ics to be mod­elled, as well as a dis­ag­gre­gated bank­ing sec­tor. A seed­ing grant to help develop ver­sion 2.0 has been recently been received from the Insti­tute for New Eco­nomic Think­ing. This will be com­bined with grants from other pri­vate enti­ties to make an appli­ca­tion for sup­port under the Aus­tralian Research Council’s Link­age pro­gram for up to A$500,000 p.a. of fur­ther fund­ing. One Aus­tralian firm has already com­mit­ted to be an Indus­try Part­ner in this appli­ca­tion, and I wel­come addi­tional sup­port from other firms, whether Aus­tralian or oth­er­wise (a min­i­mum con­tri­bu­tion of A$50,000 over 3 years is required to qual­ify as an Indus­try Part­ner under ARC rules).
  • Ver­sion 3.0 will add the capa­bil­ity to model inter­na­tional trade and finan­cial flows.

The pro­gram will be plat­form inde­pen­dent, and freely avail­able under the GPL licence.

Inte­grat­ing Min­sky with bio­phys­i­cal data

Min­sky as it stands is purely a sim­u­la­tion tool. How­ever, as part of a United Nations Envi­ron­ment Pro­gram project “Resource Effi­ciency: Eco­nom­ics and Out­look for Asia-Pacific”, a pre­cur­sor to Min­sky has been linked to a bio­phys­i­cal data­base known as ASFF (for “Aus­tralian Stocks and Flows Foun­da­tion”) devel­oped by the CSIRO (Turner, Hoff­man et al. 2011),. Our long term ambi­tion is to com­bine the two sys­tems seam­lessly, so that the phys­i­cal para­me­ters of Min­sky will be derived directly from empir­i­cal data (which can be derived for any national econ­omy) and so that Minsky’s fit to empir­i­cal data can be tested.



Fig­ure 6
The sec­ond stage of this process is part of the pro­posal for which I have just received fur­ther fund­ing from INET.

Finance and Eco­nomic Breakdown

This will be a book-length treat­ment of the Finan­cial Insta­bil­ity Hypoth­e­sis that I hope will form one of the foun­da­tions of a post-Neoclassical macro­eco­nom­ics. Writ­ing a book like this takes time and iso­la­tion, two things I have had very lit­tle of in the past six years since I first started warn­ing of an impend­ing eco­nomic cri­sis (Keen 2005). I have delayed the writ­ing of this “mag­num opus” for over a decade; in 2012–13 I intend devot­ing as much time as I can to writ­ing it, which neces­si­tates min­imis­ing time spent on other activ­i­ties such as the main­te­nance of this blog.


Cur­rently I pull in data from over 1500 dif­fer­ent sources into a Math­cad work­sheet on my PC. Math­cad, with a lit­tle help from my pro­gram­ming, does a won­der­ful job of analysing and dis­play­ing the data. But the nam­ing con­ven­tions in my pseudo-database are … a joke, there are none. Con­se­quently, only some­one inti­mately acquainted with the data can use my sys­tem, and at the moment that’s just me. I also have to man­u­ally down­load files when they are updated. Thanks to Mathcad’s vis­i­ble equa­tions, audit­ing the data is cer­tainly eas­ier than with a spread­sheet, but it is still dif­fi­cult com­pared to a well-structured rela­tional database.

A sup­porter has devel­oped an online sys­tem, cur­rently called Econ­o­data, to over­come these limitations:

  • The data is stored in a “Ruby on Rails” rela­tional database;
  • The sys­tem auto­mat­i­cally updates data when it is altered by providers;
  • The rela­tional data­base sys­tem and a 4GL for derived data series makes audit­ing straight­for­ward, and the sys­tem gen­er­ates a tinyURL so that a com­plex data series or chart can be eas­ily repli­cated by any­one; and
  • It will be eas­ily acces­si­ble and usable by sub­scribers to Debt­watch and CfESI.

Econ­o­data is cur­rently unavail­able since it is being ported to a new server, and the data­base is rel­a­tively unpop­u­lated. The data­base will also sup­port my book Finance and Eco­nomic Break­down, by mak­ing it pos­si­ble for read­ers to ver­ify any empir­i­cal charts for them­selves sim­ply by typ­ing its TinyURL into a browser.

Credit-aware Eco­nomic Indicators

My debt-aware per­spec­tive on eco­nom­ics makes it easy to explain what Bernanke has admit­ted is still inex­plic­a­ble to him: where the cri­sis came from, and why it is persisting:

“Part of the slow­down is tem­po­rary, and part of it may be longer-lasting. We do believe that growth is going to pick up going into 2012 but at a some­what slower pace than we had antic­i­pated in April. We don’t have a pre­cise read on why this slower pace of growth is per­sist­ing… ” His admis­sion of igno­rance reflects gen­uine puz­zle­ment with the economy’s fail­ure to reach what he likes to call escape veloc­ity. (G.I. 2011)

In a nut­shell, the change in total pri­vate debt is a key deter­mi­nant of aggre­gate demand, and the turn­around from increas­ing debt boost­ing demand from incomes alone by 28% in 2008 to reduc­ing demand below this level by 20 per­cent in early 2010 was the cause of the crisis.

Fig­ure 7


Sim­i­larly, the slow­down in the rate of decline of debt from its max­i­mum rate of decline of almost US$3 tril­lion p.a. to a mere $340 bil­lion p.a. is—along with the growth in gov­ern­ment debt—the main rea­son why the cri­sis has atten­u­ated slightly, rather than plung­ing into Great Depres­sion depths of unemployment.

Fig­ure 8


One indi­ca­tor that has arisen out of my work—building on orig­i­nal work by Biggs, Mayer and Pick (Biggs and Mayer 2010; Biggs, Mayer et al. 2010)—is the “Credit Accel­er­a­tor” (Keen 2011, pp. 160–165), which was first called the “Credit Impulse”. Both the change in income and the accel­er­a­tion of credit deter­mine the rate of change of eco­nomic activ­ity, and these are cor­re­lated with each other (the R2 since 1980 is 0.56), but the eco­nom­ics col­lapse in late 2007 was clearly dri­ven pri­mar­ily by the rapid and unprece­dented decel­er­a­tion of debt.

Fig­ure 9


Debt accel­er­a­tion is the main fac­tor in deter­min­ing asset prices. Asset bub­bles there­fore have to burst, because debt accel­er­a­tion can­not remain pos­i­tive forever.

Fig­ure 10


This causal rela­tion­ship is much more obvi­ous with mort­gage debt and change in house prices (see Fig­ure 11).

Fig­ure 11


Fur­ther devel­op­ment of this indi­ca­tor is there­fore highly warranted—both as an indi­ca­tor of what trends can be expected in asset prices now, and as a means to iden­tify whether a bub­ble is devel­op­ing in future. At present, the Credit Accelerator’s def­i­n­i­tion is quite simple—the change in change in debt over a time period, divided by GDP at the mid­point of that period—and the nois­i­ness of finan­cial data makes it dif­fi­cult to use short time peri­ods, which would obvi­ously be supe­rior for fore­cast­ing. A sophis­ti­cated fil­ter­ing process and for­ward indi­ca­tors for credit would make the Credit Accel­er­a­tor a much more pow­er­ful tool.