Why China Had To Crash Part 2

Flattr this!

One thing my 28 years as a card-car­ry­ing econ­o­mist have taught me is that con­ven­tional eco­nomic the­ory is the best guide to what is likely to hap­pen in the econ­omy.

Read what­ever it advises or pre­dicts, and then advise or expect the oppo­site. You (almost) can’t go wrong.

Click here to read the rest of this post.

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.
  • John­Smith

    @Prof. Keen:
    1.) Why is it that the change of debt are dri­ving the asset prices and not the total debt (plus money) in the asset mar­ket dri­ving the prices?
    When i remem­ber cor­rectly you had plot of DJ-Index (dev­ided by infla­tion) which had higher cor­re­la­tion with the total debt in com­par­i­son to change of debt.
    (And of course in a pos­i­tive feed back loop you have cor­re­la­tion between the total value and the change in value, key­word: exp-func­tion.)

    2.) Why does your pre­vi­ous cri­te­rion for the start­ing point of a finan­cial cri­sis
    (change of debt becomes big­ger than GDP)
    not work in the case of china?

  • TruthIs­ThereIs­NoTruth

    The data does not imply a causal link. You are claim­ing a causal link and the data sup­ports your claim. How­ever a bit more econo­met­ric leg­work other than cal­cu­lat­ing the cor­re­la­tion is required to sup­port the claim a bit more con­clu­sively. That is, show that the alter­na­tive hypoth­e­sis is less likely, the alter­na­tive being a) the cau­sa­tion run­ning the other way b) com­mon fac­tor c) coin­te­gra­tion. Dis­miss­ing this as a ‘smart ass’ com­ment from the crowd is sim­ply show­ing the weak­ness of your argu­ment while simul­tanously being deroga­tory with the very thing you try to asso­ciate with — math­e­mat­ics.

    Sec­ondly, you say that the best pre­dic­tor is the oppo­site of con­ven­tional the­ory. Let’s break this down, your counter claim is that mar­gin debt is caus­ing the cur­rent cycle in the Shang­hai index and that it could be used to pre­dict a burst of the bub­ble. Ques­tion is — at which moment? Once it hap­pened? Or at which point dur­ing the last two years? Does it work for every mar­ket all the time? Does con­ven­tional the­ory pre­dict the oppo­site, i.e a neg­a­tive cor­re­la­tion b/w mar­gin debt and the Shang­hai index? Or does it pre­dict no cor­re­la­tion between the two. If some part of ‘con­ven­tional’ the­ory pre­dicts no cor­re­la­tion, you haven’t exactly proven the antithe­sis, you have shown an indi­ca­tion based on one par­tic­u­lar obser­va­tion that it may not be true, but you are using econo­met­ric tech­niques to derive this cor­re­la­tion (per­haps with­out know­ing it) and not appre­ci­at­ing the prob­a­bilis­tic aspect of the sta­tis­tics, i.e. maybe there is no cor­re­la­tion and this is just a par­tic­u­larly odd obser­va­tion, until you show the like­li­hood of either you have not proven any­thing. This is a straw man argu­ment with a math­e­mat­i­cally weak antithe­sis.

  • John­Smith

    @Truth:
    I think the point Prof. Keen was mak­ing is:
    a) If the cor­re­la­tion is not zero, then the vari­ables are not inde­pen­dent (that is a math­e­mat­i­cal the­o­rem).
    (Fur­ther­more, the cor­re­la­tion is close to –1, so there is almost a lin­ear real­tion­ship between the vari­ables.)
    b) You usu­ally can’t make a causal claim just from observed (!) data.
    c) So he pro­posed a causal model (!) in terms of the graph­i­cal model imple­mented in Min­sky.
    d) These mod­els of course need to be adjusted and tested against empir­i­cal evi­dence.

  • Tim Ward

    I have never heard of a stock mar­ket bub­ble any­where that did not have a mar­gin bub­ble like­wise. If you believe that supply/demand imbal­ance causes price change, then causal­ity looks to be on fairly solid ground. (There are rea­sons not to believe at times.) Prices will not rise unless buy­ers have money in hand (exist­ing + mar­gin) to buy, and the demand volume/flow has to exceed the sup­ply volume/flow of exist­ing stock + IPO vol­ume, for prices to go up. So there has to be an accel­er­a­tion of demand beyond pre­vi­ous lev­els to see the up move. And it has to be fairly large also, because the IPO vol­ume picks up in high mar­kets. The only way this comes about empir­i­cally is by mar­gin debt cre­ation. Exist­ing money stock in the hands of retail isn’t suf­fi­cient (look at their sav­ings lev­els), because of the com­bined large vol­umes in ques­tion of exist­ing and IPOs. Mar­gin debt cre­ation is required to get the flow vol­ume up. Exam­i­na­tion of retail accounts tells you the story. Mar­gin debt is one of the pri­mary keys here. So it really does have to be causal it looks like to me, because of the need to get the demand vol­ume up sig­nif­i­cantly to get the price rise. T’other way round, how can you get the indices up, if the demand isn’t there? There are some com­pli­ca­tions, stock buy backs etc, but the basic pic­ture looks to be a fairly tight case. 

    The Modigliani-Miller the­o­rem, looks like hog-wash to me. The level of debt of a com­pany directly affects the share val­ues, by first prin­ci­ples. The shares are claims on wealth. And those claims are after the claims of deriv­a­tives and bond hold­ers. So there are sev­eral kinds of real­is­tic val­u­a­tions of cor­po­rate shares. One is the liq­ui­da­tion value. But the ele­ments of demand, futu­rity and other per­cep­tions can pro­foundly alter share val­ues from true liq­ui­da­tion val­ues. The cor­po­ra­tion could be tech­no­log­i­cally obso­lete, or in an obso­lete field, thus prac­ti­cally worth­less. Or one could be look­ing at a viable cor­po­ra­tion just at the begin­nings of one of those re-work­ings of the eco­nomic geog­ra­phy that causes huge increases in eco­nomic activ­ity and cap­i­tal. Or the cor­po­ra­tion could be engaged in account con­trol fraud. All of these will have large dif­fer­ences in val­u­a­tion from the cash flow mod­els.

    What it looks like, is the Modigliani-Miller the­o­rem is about try­ing to make a false com­par­i­son between cor­po­rate val­u­a­tion, and real estate val­u­a­tion. The cash flow/cap rate ideas make some sense in real estate. But even there, prob­lems show up. The ‘Fed model’ which is some­thing like a cap rate model, is not use­ful in my opin­ion, also for above rea­sons, and because there is no account­ing for time vari­ant risk/business cycle vari­a­tion, and pri­vate debt lev­els among other things.

  • John­Smith

    Nev­er­the­less, we need to be care­ful with causal claims.
    Pas­sive obser­va­tions are usu­ally not enough to see causal rela­tion­ships.

    The preva­lent math­e­mat­i­cal defini­ton of causal influ­ence is the “inter­ven­tional cri­te­rion of causal­ity”:

    A vari­able X has causal influ­ence on a vari­able Y, if the dis­tri­b­u­tion of the val­ues of Y changes when­ever we put/force (!) the vari­able X from a value x1 to a value x2.

    These dis­tri­b­u­tions are dif­fer­ent from the con­di­tional dis­tri­b­u­tions given X=x1 and X=x2, which are just observed (in com­par­i­son to forced on those val­ues). The dif­fer­ence is that the observed dis­tri­b­u­tions (and there­fore the sta­tis­ti­cal effect from X on Y) might be medi­ated through a com­mon cause, i.e. a third vari­able Z.
    But if we inter­vene (!) in the mech­a­nism of data gen­er­a­tion and force the vari­able X to cer­tain val­ues, then we free it from its pos­si­ble influ­ence of vari­ables Z.

  • John­Smith

    As an exam­ple, con­sider:
    X=weather
    Y=the pointer of a ther­mome­ter
    We observe a high cor­re­la­tion between sunny weather (X=sun) and high tem­per­a­ture dis­play (e.g. Y=30°). And one can’t just tell if the pointer of the ther­mome­ter is a weather con­trol­ling device or the weather caus­ing the ther­mome­ter to dis­play higher tem­per­a­ture or a higher power (Z) is con­trol­ling both.
    Only inter­ven­ing in the mech­a­nism will give us the answer:
    Putting the pointer of the ther­mome­ter to high tem­per­a­ture does not make the weather to change. So there is no causal influ­ence of the pointer onto the weather.
    But if we force the weather to be sunny in pres­ence of the ther­mome­ter (by putting the ther­mome­ter to sunny places ;-)) then we see that the pointer changes val­ues. Ergo there is a causal influ­ence of the weather on the pointer of the ther­mo­stat.
    If the last step would have also been incon­clu­sive then the “com­mon cause prin­ci­ple” would indi­cate the pres­ence of a con­found­ing vari­able Z influ­enc­ing both.

  • Tim Ward

    Those are inter­est­ing points, John.

    Grap­pling with the prob­lem, requires some clear def­i­n­i­tions, for exam­ple just what is price? As dis­tinct from bid and offer, which are con­di­tional, and can be with­drawn. Price is defined by the last trans­ac­tion. So when one bases analy­sis on actual trans­ac­tions, the causal­ity would seem to become clear. There had to be actual money avail­able to trans­act. Where did that money come from? Was it already in exis­tence? Or was it newly cre­ated for the trans­ac­tion (mar­gin)? (Or some mix­ture.) When exist­ing money for trans­ac­tion has been exhausted, only new money, mar­gin, can answer. So when one looks at what is going in terms of the volumes/flow of orders and trans­ac­tions, it becomes clear that the amount of exist­ing money avail­able for trad­ing, in trad­ing accounts, can­not sup­port the kind of price rises we see in bub­bles, with­out the addi­tional mar­gin cre­ated. Think about the amount of trans­ac­tions that would occur if the mar­gin trans­ac­tions did not occur. The demand and vol­umes would be smaller, and the surge required for price rises in bub­bles would not occur,( in part because the IPOs soak up so much money). The index lev­els are deter­mined by actual trans­ac­tions, and thus the money put up for the trans­ac­tions and what is going on with vol­umes. This gives us a causal link. At least that’s the way it looks to me. 

    So it looks like your inter­ven­tion con­di­tion is actu­ally right there, the mar­gin debt/credit is the inter­ven­tion, and if one with­drew the inter­ven­tion, the vol­umes would be lower, and one could not get the surge required to get the price rises. Or if you pre­fer, inter­ven­ing to pre­vent mar­gin increases, would lower demand and vol­umes with cor­re­spond­ingly lower price lev­els. And we actu­ally have evi­dence on that topic as well. In arti­cles from var­i­ous news out­lets, when the CME (or other mar­kets) alter mar­gin require­ments, by increas­ing mar­gin require­ments, it results in price decreases. This occurred in the sil­ver mar­ket and oth­ers, it’s some­thing I’ve been pay­ing atten­tion to. When mar­gin require­ments are increased (more equity required), prices drop. It affects actual cur­rent posi­tion­ing. When mar­gin require­ments are relaxed, prices can go up, but they don’t have to, it affects future deci­sions.

    So, thanks for bring­ing up the point about inter­ven­tion, because we actu­ally have sup­port­ive evi­dence there! Increas­ing mar­gin require­ments low­ers price lev­els.

    A sim­i­lar thing occurs in hous­ing mar­kets, its in the loan writ­ing stan­dards. When stan­dards are tight­ened (larger down pay­ment and higher income lev­els), it puts a damper on sales, and prices. Loos­en­ing lend­ing stan­dards (‘increas­ing mar­gin’) is asso­ci­ated with bub­bles. So the mar­kets behave sim­i­larly.

    (And on the other topic, another val­u­a­tion method con­trast­ing to the Modigliani-Miller the­o­rem would be the replace­ment cost val­u­a­tion.)

  • Tim Ward

    It just occurred to me that the inter­ven­tion prin­ci­ple destroys the main­stream posi­tion the money doesn’t mat­ter, it’s just a veil over trans­ac­tions.

    If money were just a veil, then changes in the lev­els of trans­ac­tions by alter­ing mar­gin require­ments (and loan writ­ing stan­dards) would not be observed. 

    But such a proof is not really required, its obvi­ous that money and credit cre­ation has a rul­ing, causal func­tion. Proof by inspec­tion?

  • John­Smith

    @Tim: Your analy­sis seems to be cor­rect!
    If we fixed mar­gin at a low level then the trans­ac­tion volume/price/numbers would be less in com­par­i­son to the case we fixed mar­gin at a high level.
    So there seems to be a causal influ­ence of mar­gin to trans­ac­tion volume/price/numbers.

  • Tim Ward

    The inter­ven­tion prin­ci­ple also applies to inter­est rates and FX. Changes in inter­est rates and FX affect sales and vol­umes.

  • TruthIs­ThereIs­NoTruth

    You seem to be say­ing that mar­gin lend­ing is the exclu­sive force by which trans­ac­tions occur. Firstly note that trans­ac­tions occur when prices are ris­ing or falling, the exis­tence of trans­ac­tions doesn’t imply ris­ing prices. Sec­ondly, prices can rise with or with­out mar­gin lend­ing, to say that prices can rise if and only if mar­gin lend­ing rises is a very strong state­ment which is both non intu­itive and with­out even hav­ing done the work can guar­an­tee is empir­i­cally false. An alter­na­tive hypoth­e­sis is that a ris­ing mar­ket induces more risk tak­ing and lever­age which has to unwind through mar­gin calls, I have seen anec­do­tal evi­dence to this effect. Show­ing which hypoth­e­sis is more likely requires more than art­ful words. This is the prob­lem with econ­o­mists, they think they can replace proof with essays, which is fine but to call your­self a math­e­mati­cian or say that you take a math­e­mat­i­cal approach you need to respect the rules of the game oth­er­wise it is sim­ply a fraud­u­lent claim. 

    I’m actu­ally not say­ing it is one way or the other. A key sta­tis­tic is how many of the trans­ac­tions are actu­ally per­formed with mar­gin lend­ing. I have seen many cases where mar­gin lend­ing is sim­ply fol­low­ing the mar­ket, it becomes pop­u­lar when the mar­ket is ris­ing and is unwound when the mar­ket is falling but over­all forms an uni­flu­en­tial pro­por­tion of trans­ac­tion. That is not to say that it can’t become an influ­ence if the vol­ume is large enough. But it takes more than bald state­ments to show which one it is for a par­tic­u­lar cir­cum­stance. To not only make bald state­ment but then to extrap­o­late that state­ment and fur­ther still use it to broadly sup­port your approach by turn­ing a 60 year old the­ory into a straw man for your lin­guis­tic bay­o­nets is either naive or prey­ing on a naive audi­ence.

  • John­Smith

    @Truth:
    There can be more than one vari­ables hav­ing causal influ­ence on a given vari­able. For exam­ple there are dif­fer­ent rea­sons you can get a headache, e.g. a cold or drink­ing or a bar fight.
    Fur­ther­more, there also could be causal feed­back loops.
    For exam­ple the size of the pop­u­la­tion has an influ­ence on the num­ber of chil­dren born, which in return has influ­ence on the size of pop­u­la­tion (pos­i­tive feed­back loop).
    Or e.g. fox pop­u­la­tion and rab­bit pop­u­la­tion, where foxes eat rab­bits (neg­a­tive feed­back loop).

  • Tim Ward

    TruthIs­ThereIs­NoTruth, “You seem to be say­ing that mar­gin lend­ing is the exclu­sive force by which trans­ac­tions occur.” Ha ha ha! I’ve done noth­ing of the sort. An explicit state­ment about exist­ing money also appears. We were explic­itly talk­ing about bub­bles, and what is going on there. I haven’t made any state­ment that what I’ve writ­ten is a for­mal proof. Nor, ‘QED’ or any­thing like that. I said basi­cally, that it looks to me that there is a causal link, and con­sid­er­a­tions in form­ing that opin­ion. I didn’t say any­one had to buy it. To me it cer­tainly looks that there is a causal link between mar­gin debt and bub­bles, and you haven’t said any­thing to alter that opin­ion. We could talk also about huge sell-offs, but we weren’t at the time.

    I could go on and answer other points, but I won’t because it is evi­dent by inspec­tion the nature of your dis­cus­sion, as I have seen on pre­vi­ous posts. There really are causal rea­sons for cer­tain eco­nomic and mar­ket phe­nom­ena, and some peo­ple that don’t want any­one to know about it. Maybe you’re one of them. 

    You’re wel­come to the to the Modigliani-Miller the­o­rem, if you like it, and the cap rate meth­ods. I sim­ply said, to para­phrase, that in my opin­ion, it was not cor­rect, and gave rea­sons why, includ­ing other val­u­a­tion con­sid­er­a­tions. And I have some other time vari­ant val­u­a­tion con­sid­er­a­tions, and have read some other inter­est­ing dis­cus­sions about cash flow val­u­a­tions. But they will obvi­ously stay pri­vate now. You can trash those also if you wish. You needn’t lis­ten to a thing I say, you can change the chan­nel. Each to his own. And you can keep that veil over trans­ac­tions thingy if you want that too, and equi­lib­ria, the DGSE stuff, and mar­ginal util­ity, and a bunch of other main­stream stuff if you like it. 

    Some peo­ple are inter­ested to find out if there are any other ways of look­ing at eco­nom­ics. And there are.

    Cap­i­tal­ism is sales. You’ll meet a good sales­man.” Who said that?