Deleveraging, Deceleration and the Double Dip

• Tweet

• Tweet

Much opti­mism flowed from last week’s dec­la­ra­tion by the Nation­al Bureau of Eco­nom­ic Research that the US reces­sion offi­cial­ly end­ed in June 2009. How nice of them to let us know.

Mar­kets react­ed warm­ly and the 8 per cent ral­ly in US stocks through Sep­tem­ber seemed more impor­tant than the rev­e­la­tion that the US Fed is wor­ried enough about defla­tion to be plan­ning anoth­er round of quan­ti­ta­tive eas­ing — dubbed ‘QE 2’.

I wish I could share the mar­ket’s (and the NBER’s) irra­tional exu­ber­ance, but the key indi­ca­tor that explains where the US econ­o­my and its already dis­as­trous employ­ment sit­u­a­tion is head­ed implies that even QE2 won’t set the US on a course to renewed pros­per­i­ty.

It’s also impor­tant to note that the ‘strong’ Q2 US earn­ings fig­ures seen in July are part­ly the result of dra­mat­ic cost-cut­ting in US firms – a nice way of say­ing mass lay-offs. Nonethe­less the stronger bot­tom lines keep pro­duc­ing mar­ket exu­ber­ance, even if the fac­tors behind those bot­tom lines will lead to future defla­tion rather than a boom. As Forbes writer Joshua Brown puts it “there is an effer­ves­cence in the air as we head into the Q3 report­ing peri­od (start­ing Octo­ber 7)”. He thinks the mar­ket will once again turn south, and based on that pesky fun­da­men­tal called the real econ­o­my, it should.

So should it on the basis of the key indi­ca­tor that explains the ori­gins of the appar­ent sta­bil­i­sa­tion that led the NBER to declare that the reces­sion was over in June 2009.

For a long time I’ve focused on the con­tri­bu­tion that the change in debt makes to aggre­gate demand, in the rela­tion that “aggre­gate demand equals the sum of GDP plus the change in debt”. An obvi­ous exten­sion of that was that “change in aggre­gate demand equals change in GDP plus accel­er­a­tion in the lev­el of debt”—which would imply that change in unem­ploy­ment is dri­ven by changes in the rate of growth of debt.

Though I was aware of this impli­ca­tion of my analy­sis, I held off from test­ing it because I was con­cerned that this was push­ing the data one step too far.

A phys­i­cal sys­tem with a sim­i­lar rela­tion­ship between veloc­i­ty (the rate of change of one vari­able) and accel­er­a­tion (whether the veloc­i­ty of anoth­er vari­able is increas­ing or decreas­ing) would gen­er­ate a large vol­ume of suf­fi­cient­ly detailed data that the rela­tion­ship could be empir­i­cal­ly test­ed.

But the eco­nom­ic sys­tem, with the large time lags in data col­lec­tion, sur­vey meth­ods rather than direct mea­sure­ment, and the dodgy prac­tices sta­tis­ti­cians are forced into by politi­cians and eco­nom­ic bureau­crats who often don’t want raw infor­ma­tion to be avail­able? I just thought that the rela­tion­ship, even though it made sense, would­n’t be dis­cernible from pub­lished sta­tis­tics. So I held off.

It turns out that I should­n’t have been so cau­tious: the data well and tru­ly sup­ports this, on the sur­face, weird causal rela­tion: the change in employ­ment is strong­ly affect­ed by the accel­er­a­tion or decel­er­a­tion of debt. This can give the para­dox­i­cal result that the lev­el of employ­ment can rise, even when the econ­o­my is delever­ag­ing, if the rate of delever­ag­ing slows. This phe­nom­e­non has dri­ven the appar­ent sta­bil­i­sa­tion of the US unem­ploy­ment rate (though of course the more mean­ing­ful U‑6 mea­sure has risen to 17 per­cent, and Shad­ow­stats puts the actu­al unem­ploy­ment lev­el at 22.5 per­cent–well and tru­ly in Depres­sion ter­ri­to­ry), and it is high­ly unlike­ly that it will last.

My unchar­ac­ter­is­tic timid­i­ty means that I have to doff my cap in the direc­tion of the three econ­o­mists who first pub­lished on this top­ic: Big­gs, May­er and Pick. They first showed the cor­re­la­tion between what they called “the cred­it impulse”—the rate of change of the rate of change of debt, divid­ed by GDP—and both GDP and employ­ment (for those who have access to research from Deutsche Secu­ri­ties, they have a sim­pler expla­na­tion of their analy­sis in Glob­al Macro Issues for Decem­ber 17 2009: “The myth of the cred­it-less recov­ery”).

The chart below shows my con­fir­ma­tion of the rela­tion­ship with the data on the annu­al change in unem­ploy­ment in the USA and the annu­al rate of accel­er­a­tion of pri­vate debt since 1955. The cor­re­la­tion is ‑0.67: a stag­ger­ing cor­re­la­tion of a first and a sec­ond order vari­able over such a peri­od, and across both booms and busts.

The two dot­ted red lines labelled “S” and “E” show when the NBER thinks this reces­sion start­ed and end­ed, and they neat­ly coin­cide with turn­ing points in the cred­it impulse—an indi­ca­tor that the NBER is not even aware of, let alone one that it con­sid­ers when attempt­ing to date reces­sions.

Super­fi­cial­ly, one might think that since the cred­it impulse does indi­cate when unem­ploy­ment is going to rise or fall, then the cur­rent data implies that the reces­sion is indeed over—even if the NBER does­n’t under­stand the actu­al causal dynam­ics at play.

But the chart also shows that there has nev­er been a turn­down in cred­it like this one—the peak rate of decel­er­a­tion of debt was over 25 per­cent, ver­sus a mere minus 6 per­cent in the deep reces­sion of the 1970s. And though the rate of accel­er­a­tion of debt has the most direct impact on employ­ment, ulti­mate­ly all three factors—the lev­el of debt (com­pared to GDP), its rate of change, and whether that rate of change is increas­ing or decreasing—must be tak­en into account.

It’s com­pli­cat­ed, so an anal­o­gy with dri­ving makes it eas­i­er to com­pre­hend.

Con­sid­er a dri­ve from Los Ange­les to some des­ti­na­tion East (if you’re an Aus­tralian read­er, con­sid­er a dri­ve West from Syd­ney), where the dri­ve out rep­re­sents increas­ing debt, and the dri­ve back home rep­re­sents falling debt.

The lev­el of debt com­pared to GDP is like the dis­tance to be trav­elled, and today the US has a lot fur­ther to trav­el than it did in the 1950s: 5 times as far, in fact. It’s like the dif­fer­ence between a dri­ve to New York and back, ver­sus a return trip to Utah.

The rate of change of debt (with respect to GDP) is like your speed of travel—the faster you dri­ve, the soon­er you’ll get there—but there’s a twist. On the way up, increas­ing debt makes the jour­ney more pleasant—the addi­tion­al spend­ing increas­es aggre­gate demand—and this expe­ri­ence is what fooled neo­clas­si­cal econ­o­mists (who ignore the role of debt) into believ­ing in “the Great Mod­er­a­tion”. But it increas­es the dis­tance you have to trav­el when you want to reduce debt, which is what the USA is now doing. So it’s great when you’re dri­ving from LA East (increas­ing debt), but lousy when you want to head home again (and reduce debt).

With that far to trav­el back home, you might be tempt­ed to accelerate—which is akin to increas­ing the rate of change of the rate of change of debt (it’s a mea­sure of the g‑forces, so to speak, when they can be gen­er­at­ed by either rapid accel­er­a­tion or rapid decel­er­a­tion). Accel­er­a­tion in the debt lev­el when it was ris­ing again felt great on the way out: booms in the Ponzi Econ­o­my the US has become were dri­ven by accel­er­a­tions in the rate of growth of debt. Equal­ly, accel­er­a­tion in the oppo­site direc­tion feels dread­ful: as the rate of decline of debt increas­es, aggre­gate demand col­laps­es and unem­ploy­ment explodes.

What actu­al­ly feels bet­ter in the reverse direc­tion is deceleration—reducing the rate at which debt is falling—and that’s what’s been hap­pen­ing in the last year.

But here’s the prob­lem: too much decel­er­a­tion and you actu­al­ly reverse direc­tion: you start head­ing East again, rather than return­ing home. That would­n’t be a prob­lem if all you’d done was dri­ve to Utah, but instead you’ve hit New York instead: dri­ve any fur­ther, and you’re in the Atlantic.

With the lev­el of debt the USA has accu­mu­lat­ed, the prospect that any sec­tor of it (apart from the gov­ern­ment) can be enticed to go back into accu­mu­lat­ing debt once again is remote. So the decel­er­a­tion in the rate of reduc­tion of debt that is occur­ring right now will ulti­mate­ly give way to at best a con­stant rate of decline of debt, and at worst anoth­er acceleration—and the dread­ed “dou­ble dip”.

These next two quar­ter­ly charts empha­sise the dilem­ma: the sta­bi­liza­tion in employ­ment has occurred because the rate of delever­ag­ing has slowed, which reg­is­ters as a pos­i­tive in the “rate of change of the rate of change”:

But the rate of change of debt is still neg­a­tive: it’s just risen from a low of ‑6% to ‑2%. For the decel­er­a­tion effect to con­tin­ue, the US debt car would need to stop its return dri­ve from New York to LA, and head back towards New York once more: the lev­el of debt rel­a­tive to GDP would need to rise.

This is high­ly unlike­ly when all sec­tors of the Amer­i­can econ­o­my bar one (non-finan­cial busi­ness) are already car­ry­ing more debt than they were in the depths of the Great Depres­sion, when the debt ratio had been dri­ven high­er by defla­tion.

So the decel­er­a­tion in delever­ag­ing should give way again at some point, and then the NBER may be forced to begin dat­ing the next recession—which is still a con­tin­u­a­tion of the cur­rent Depres­sion.