An interesting challenge

• Tweet

• Tweet

There’s an inter­est­ing post on the Atlantic by Jim Manzi, Stim­u­lus pre­dic­tions: put up or shut up, that calls on econ­o­mists who are mak­ing pre­dic­tions about what Oba­ma’s stim­u­lus pack­age will or won’t do to present their mod­els on which these pre­dic­tions are based.

In part, he says:

So here’s what we would need to fal­si­fy a pre­dic­tion.  Any­one who claims to know the impact should escrow a copy of the source code of the econo­met­ric mod­el that is used to make the pre­dic­tion, along with a stat­ed con­fi­dence inter­val, oper­a­tional scripts, and assump­tions for all required non-stim­u­lus inputs that pop­u­late the mod­el with a named third-par­ty.  Upon reach­ing the date for which the pre­dic­tion is made, the third-par­ty should run the mod­el with the actu­al data for all non-stim­u­lus assump­tions and com­pare the mod­el result to actu­al.  Any dif­fer­ence would be due to mod­el error.  We actu­al­ly still would not be able to par­ti­tion the sources of error between “error in pre­dict­ing causal impact of stim­u­lus” and “oth­er”, but at least we would have a real mea­sure­ment of mod­el accu­ra­cy for this instance.

Of course, I sin­cere­ly doubt this will hap­pen.  I won­der why not?

As read­ers of this blog will under­stand, I don’t make empir­i­cal pre­dic­tions from my mod­els, but I do make qual­i­ta­tive ones; and my mod­els aren’t the econo­met­ric giants that con­ven­tion­al econ­o­mists build, but small­er mod­els that, in con­trast to the econo­met­ric lot, are tru­ly dynam­ic (for those who don’t realise that stan­dard eco­nom­ic mod­els aren’t gen­uine­ly dynam­ic, please read this blog post “Why Did I See it Com­ing and “They” Didn’t?” ).

So I’ve tak­en up Jim’s chal­lenge, and wrote the fol­low­ing com­ment on his blog:

Dear Jim,

I’m will­ing to take your chal­lenge, but from an unusu­al per­spec­tive.

While I am an aca­d­e­m­ic econ­o­mist, I don’t build nor believe in the type of econo­met­ric mod­els that dom­i­nate eco­nom­ics these days–generally so-called “New Key­ne­sian” or “Dynam­ic Sto­chas­tic Gen­er­al Equi­lib­ri­um” mod­els.

Instead I build non­lin­ear dynam­ic mod­els based on Min­sky’s “Finan­cial Insta­bil­i­ty Hypoth­e­sis”, and I have start­ed con­struct­ing a strict­ly mon­e­tary mod­el of a pure cred­it econ­o­my.

My pre­dic­tions based on these mod­els are qual­i­ta­tive rather than quan­ti­ta­tive, but on the grounds of Min­sky’s extreme­ly pre­scient hypoth­e­sis the sheer scale of pri­vate debt that has been accu­mu­lat­ed, and the abun­dant his­tor­i­cal data on debt with which we can review past eco­nom­ic per­for­mance in the light of Min­sky’s hypoth­e­sis, I have been argu­ing that this cri­sis is beyond bailouts.

There­fore while I think the bailouts are bet­ter than doing noth­ing, ulti­mate­ly I see them as futile. All they will do is replace some pri­vate debt with even more pub­lic debt as has hap­pened in Japan (if the spend­ing is debt-financed), or pump fiat mon­ey into the econ­o­my only to see it dis­ap­pear into debt repay­ment and not reflate the econ­o­my if (as Bernanke is now doing with M0) the heli­copter approach is used.

To check my rea­son­ing and qual­i­ta­tive pre­dic­tions on this front, I prof­fer two mod­els that are elu­ci­dat­ed on posts on my blog http://www.debtdeflation.com/blogs:

http://www.debtdeflation.com/blogs/2008/11/26/parliamentary-library-vital-issues-seminar/

and

http://www.debtdeflation.com/blogs/2009/01/31/therovingcavaliersofcredit/

The for­mer includes a mod­el of Min­sky’s Hypoth­e­sis built in the sys­tems engi­neer­ing pro­gram Vis­sim, which is down­load­able from the blog; the lat­ter details a mod­el of a pure cred­it econ­o­my that under­goes a cred­it crunch, using the math­e­mat­ics pro­gram Math­cad.

The for­mer can be down­loaded and run; the lat­ter I only explain in the post, though there is a draft of a forth­com­ing paper that details the math­e­mat­ics of the mod­el:

http://www.debtdeflation.com/blogs/wp-content/uploads/papers/NotKeenOnBailoutsFinal.pdf

Both mod­els, which I’m now work­ing on inte­grat­ing and extend­ing, imply that there is no way out of this cri­sis while we still in effect hon­our the debt that was run up dur­ing this spec­u­la­tive bub­ble. We either have to inflate it out of exis­tence, or selec­tive­ly abol­ish it.

Since, as you’ll see from the sec­ond post above, I also doubt the pos­si­bil­i­ty of caus­ing infla­tion sim­ply by dri­ving up M0, the sec­ond option is the only one that I expect will work: at some point, to end this cri­sis, much of the debt is going to have to be repu­di­at­ed.

I’ll keep an eye on that blog entry to see what even­tu­ates, and whether any neo­clas­si­cal econ­o­mists sub­mit their mod­els for scruti­ny.