The failure of creditism

Posted by Tim Congdon in News Archive | 0 comments

A recurrent theme of media commentary and punditry in recent years is that ‘extra bank lending’ (in some sense, usually left vague) is a condition of wider macroeconomic recovery. This contention tends to be the prelude to another round of bank-bashing, with politicians and others blaming banks’ actual or alleged reluctance to lend as the reason for the sluggishness of demand, high unemployment and so on. The whole approach depends on the existence of a relationship between bank lending to the private sector and aggregate expenditure (or perhaps some subset of aggregate expenditure with an assumed powerful influence over the total). The claim that there is a relationship of this sort has no place in standard monetary theory, but has become  fashionable  in  the  last  20  years  or  so  largely because  of  papers written by academics from the leading universities on the USA’s East Coast (Harvard, Princeton, Columbia, New York University and so on). Arguably, the current chairman of the Federal Reserve, Ben Bernanke, is the principal exponent of ‘creditism’, the set of ideas that pivots on the lending-expenditure relationship.

The  current  weekly  e-mail  note  is  the  first  of  a  series  that  will  criticize ‘creditism’. (In my opinion they will demolish it, but others must decide for themselves.)  My focus  here  will  be  on the  UK  in  the  period  of  the  Great Recession, but I believe the arguments to have wider validity. At any rate, I will show both that no relationship holds between bank lending to the private sector and nominal GDP in the UK in recent years, and that there are good reasons why a relationship between the two variables is not necessarily to be expected. (I do not deny that, in the normal course of events, new bank lending creates new bank deposits, which are money, and that a relationship holds between the quantity of money and nominal GDP. But it is the quantity of money as such, not bank lending, that is doing the vital work in the determination of nominal GDP.)

The meaning of ‘creditism’

A recurrent theme of media commentary and punditry in recent years is that ‘extra bank lending’ is a condition of wider macroeconomic recovery. The approach turns on the supposed ‘special’ nature of bank lending in the macroeconomic firmament, and was termed ‘creditism’ in a 1995 academic paper by Ben Bernanke and Mark Gertler. For the ideas to be cogent, the creditist school has to posit the existence of a relationship between bank lending to the private sector and aggregate expenditure (or an important subset of aggregate expenditure). Many of the leading papers are less than specific about how bank lending is supposed to affect spending, but their authors do not seem to regard the lack of clarity as a problem. In their view, a credit-based analysis is a rival to money-based analysis, and a money-based analysis has similar or worse deficiencies. (Bernanke and Gertler have in fact characterized the relationship between money and expenditure as ‘a black box’. This is weird, as numerous accounts of this relationship [and of the implied transmission mechanism from money to the economy] are to be found in a classic literature going back to David Hume in the 18th century. I regard the ‘black box’ allegation as poppycock. I am afraid that Bernanke and his associates simply have not read enough, or perhaps have not understood what they have read. In my own work see, for example, essays 15 and 16 in my 2011 book Money in a Free Society as well as the preface [on pp.325 – 9] to the final part of that book.)

Several hypotheses on the lending/expenditure relationship are to be found in the creditist literature, often only implicitly. I suggest below five such hypotheses, although it should be emphasized that this list is not meant to exclude other related ideas.

  1. The level of aggregate nominal expenditure ‘depends on’ (or ‘is a function of’) the level of bank lending to the private sector. (The level of bank lending means ‘the stock of loans’ or ‘the outstanding total’ of loans; it does not mean the flow of new bank lending.)
  2. The level of aggregate nominal expenditure depends on the change in bank lending to the private sector. (The change in bank lending to the private sector is the flow of new lending, i.e., the addition or subtraction to the stock of loans.)
  3. The change in aggregate nominal expenditure depends on the change in bank lending to the private sector. (This might appear to be implicit in the first relationship, but testing for first differences is more rigorous than testing for levels.)
  4. The level of aggregate nominal expenditure depends on ‘credit spreads’, where ‘a credit spread’ is the excess of the borrowing rate on a bank loan over the banks’ cost of funds or some other bellwether interest rate on a very safe security or loan, and where the credit spread impacts on demand partly via its effect on bank lending (and hence at a further remove on aggregate nominal expenditure) and partly via other effects. (Some statements seem to mean that credit spreads affect the change in expenditure, not the level. I am not going to bother with this notion, which is plainly pretty silly. The lack of testable hypotheses in many of the creditist papers is in fact a disgrace.)
  5. The level of aggregate nominal expenditure depends on a ‘credit breakdown’ (whatever that is), which in turn is to be blamed on ‘information asymmetries’. Information asymmetries are systemic in capitalist financial systems, including the banks, and symptomize the inefficiency of banks, finance, capitalism etc.

The first of these five propositions is to be found in papers from Ben Friedman (not Milton Friedman, please note) of Harvard University; the second is a commonplace of low-level newspaper punditry (‘the banks must clean up their balance sheets so that they can lend more and spark a recovery’ etc.), but is also found in IMF material; the third proposition is not independent from the first; the fourth is contained in papers from Bernanke, Gertler and others, typically dressed up with references to an assortment of ‘channels’; and the fifth originates in papers by the Nobel-prize-winning economists Jo Stiglitz (of Columbia University) and George Akerlof (of the University of California, Berkeley). There is no room in this particular note to look into the fourth and fifth propositions in any detail. The focus will be on the relationships in the UK’s Great Recession between

  1. The level of bank lending to the private sector and the level of nominal GDP,
  2. The change in bank lending to the private sector and the change in nominal GDP
  3. The change in bank lending to the private sector and the level of nominal GD

It will be found that none of these relationships holds water. They are all worthless, according to the usual statistical tests. After looking at the data, an argument for the lack of a relationship between bank lending and national income will be developed.

Before getting into the numbers, it should be noted that the figures for ‘bank lending to the private sector’ in the current exercise exclude lending to so-called ‘intermediate other financial corporations’. Inter-bank lending and deposits are invariably kept out of analyses of the present type, as banks do not account for much of aggregate demand and their own ‘expenditure’ (in the Keynesian sense) is not related to their deposits or loans in the same way as with non-bank agents.Were bank  lending and GDP related in the  UK's latest boom-bust cycle?

The levels of bank lending and nominal GDP

The main item of evidence is presented in the chart above, which relates to the period from the start of 2008 (commonly viewed as the start of the UK’s Great Recession) to the second quarter of 2012. The levels of lending and nominal GDP moved in opposite directions for most of this period. In the year to Q1 2009 lending rose by 3.6%, whereas money GDP dropped by 4.5%. By contrast, in the two years to Q1 2011 lending dropped by 14.8%, whereas money GDP advanced by 8.4%. The fall in bank credit to the private sector in the two years to the start of 2011 was the sharpest in the post-war period, but it did not prevent the economy recovering in 2010 and 2011 from the severe downturn in 2009. From early 2011 the level of bank lending to the private sector stagnated, while nominal GDP continued to make modest gains. (For those who are interested, when the level of nominal  GDP was regressed on the level of lending, the two series were reasonably correlated, but that is not unusual in series for levels. More important, the regression coefficient was not significantly different from zero. This destroys the idea that lending mattered in the causation of expenditure and income.)Were bank lending and GDP related in the UK's latest boom-bust  cycle

Changes in bank lending and nominal GDP

More interesting as a test of the creditist hypothesis is to look at the relationship between changes in bank lending and nominal GDP. A relevant chart is shown above, although it relates to the period of both ‘boom’ and bust in the latest cycle. The three-and-a-half-year period to the end of 2007, which were mostly of buoyant asset prices and above-trend growth (i.e., of ‘boom’) are included, as well as the Great Recession from the start of 2008. Visual inspection is far from encouraging from a creditist point of view, even if the years of the ‘boom’, such as it was, saw simultaneous increases in both lending and nominal GDP. From the start of 2008 the changes in lending and nominal GDP were in opposite directions most of the time. More precisely, the values of the six-month change in lending were negative in the eight quarters from 2009 Q2, whereas over the same period they were positive for the six-month change in nominal GDP. More careful econometric work fails to identify any worthwhile relationship for the creditist school. The most damning feature of this work is that in the Great Recession itself the regression coefficient in the lending/GDP relationship was negative. In other words, a reduction in lending was associated with an increase in GDP and, on the face of it, an increase in lending with a reduction in GDP! The whole line of thought is tripe.

(If the change in nominal GDP is regressed on the change in lending, the resulting equation for the data in the eight years to Q2 2012 is,

 Change in nominal GDP [% annualized rate] = 2.93 + 0.12 Change in bank lending [% annualized rate],

with a r2 of 0.074 and the t statistic on the regression coefficient of 1.55. The equation for the data in the four and a half years to Q 2012, i.e., the period more specifically of the Great Recession, is

Change in nominal GDP [% annualized rate] = 0.96 – 0.32 Change in bank lending [% annualized rate],

with a more promising r2 of 0.218, but a negative regression coefficient just on the borderlines of significance [with a t statistic of 2.11].)

Some simple theory: Why the level of bank lending has, by itself, no bearing on the level of nominal GDP

As already noted, the assertion that ‘national income is a function of bank lending to the private sector’ has no place in standard theory, and has only come to prominence in the last 20 years or so because of papers from Ben Friedman and others. The standard theory is that national income is a function  of  the  quantity  of  money,  with  the  key  proposition  being  that  national  income  is  in equilibrium only when the demand to hold money balances is equal to the quantity of money created by the banking system. (See essay 11 in my 2011 Money in a Free Society.) Unfortunately, the standard theory suffers from several important ambiguities. These ambiguities are partly attributable to the development of the standard theory from the so-called ‘quantity equation’, MV = PT, where M is defined as the quantity of money, V as the velocity of circulation, P is the price level and T is the number of transactions.

As has been often remarked (and was fully understood by Irving Fisher when he  proposed the quantity equation), the symbols in the MV = PT identity could refer to several concepts. An obvious temptation is to regard T as synonymous with national income (and output and expenditure, as in textbook Keynesianism) and to replace T with Y. We then have MV* = PY, where V* is the ‘income velocity of money’. The phrase ‘income velocity of money’ is indeed so familiar that, for many analysts, it is seen as equivalent to the ‘velocity of money’. The velocity of money is then a variable that ties together the quantity of money and national income (or output or expenditure), and money is seen as having its principal macroeconomic significance in the determination of national income. The price term (i.e., P) can be understood as approximated by the GDP deflator. If both sides of the identity are deflated by P, the expression starts to look like a behavioural equation, with real money balances linked to real incomes  and V*  open to interpretation with the tools of money demand function analysis.

This is fair enough, and in many contexts the tendency to see the income velocity of money as ‘the’ velocity of money is convenient and helpful. However, there is a very different way of looking at the subject. In a modern economy the value of payments is a high multiple (50 or so times in the UK’s case) of the value of national income. The high multiple to national income reflects

  1. The use  of  money  in  income  payments  (such  as  payments  of  wages  and  salaries), expenditure   payments   (in,   for   example,   the   retail   sales   that   constitute   part   of consumption) and output payments (when purchasing goods from a factory, where the value is added),
  2. The use of money to cover so-called ‘intermediate transactions’, where the payment to the seller is larger than (and sometimes a multiple of) the value added by the seller. (A retailer pays a wholesaler a sum of money which reflects the wholesaler’s margin, the wholesaler’s own costs and the payment made to reflect the value added at an earlier stage of supply.)
  3. The use of money to expedite the transfer of incomes between several agents, with, for example, tax payments to the state, which pays social security benefits, which are taxed, etc., so that payments are higher than the income which is part of ‘national income’,
  4.  The use  of  money  to  facilitate  the  shift  of  balances  between  accounts  in  the  same beneficial ownership, including payments between companies’ subsidiaries, between individuals’ savings deposits and their current accounts, and so on, and, finally (but very far from least),
  5. The use of money in transactions in capital assets, that is, in the purchase and sale of corporate equity, real estate, bonds and so on

We could restate the quantity identity as follows, MV# = ‘P’.T, where V# becoming the ‘turnover velocity of money’, ‘P’ the price level of some composite of the goods and services in the income- expenditure flow and of capital assets (and what a monster that price index would be and how necessary are the quotation marks around the P!) and T as the number of payments. It perhaps hardly needs to be said that the number of transactions in a modern economy would run into the trillions. As far as I am aware, no exact quantification of this supposed ‘T’ has ever been attempted. (An inevitable observation here is that the MV = PT formula is far trickier than it looks, and is riddled with ambiguity and complexity. The MV = PT approach turns out to be something of a dead-end. Indeed, it was precisely these ambiguities and complexities which encouraged Cambridge economists in the 1920s and 1930s to prefer a different statement of the underlying relationships. Their reformulation – so that equilibrium was defined as the equivalence of the demand to hold money with the quantity of money balances actually in existence – was deliberately intended to marry up economy-wide equilibrium with the Marshallian demand-equals-supply equilibria of particular companies and industries. The reformulation was the work particularly of Pigou, Robertson and Keynes.)

My point is that – if we think about total payments and see money as linked to the value of total payments thought the turnover velocity of money – we quickly see why creditism is rubbish. The total value of payments in the UK today is, roughly, £75,000 billion a year. (Yes, £75,000 billion, not £75,000 million, which would of course be £75 billion.) All those payments are made with money, predominantly  nowadays  by  electronic  payments  across  bank  accounts,  with  the  ‘same’  money moving around between the accounts dozens of times a year.

Now let us think about bank lending to the private sector. The box below shows the values of the change in the stock of UK bank lending in recent years. As is well-known, banks were expanding their loan portfolios aggressively – too aggressively – in in the three years to 2007, with the total amount of net new lending in the period being almost £600b., equivalent to about half of GDP at the time. When stated that way, with new lending over a three-year period being equal to half of annual Change in stock of UK bank lending

GDP, it seems obvious that the lending – either the level or the change – is relevant to the determination of national income. (Well, relevant somehow or other, never explained very clearly.) But – if we compare the lending numbers with the value of all payments in the economy – the notion that lending matters to the value of payments falls apart. How can numbers like £235.8 billion or minus £51.4 billion have any bearing on £75,000 billion? Let us remember that all the transactions that constitute the £75,000 billion are transactions that require settlement in money. The equilibrium value of payments can sensibly be interpreted as the quantity of money multiplied by the desired turnover velocity of money, with bank lending by itself  not of any relevance to the story. The Ben Friedman claim that the level of bank lending is related to national income is shown to be wrong, hopelessly wrong. Agents turn over £75,000 billion a year because they have the quantity of money that enables them to complete that value of transactions. It is the quantity of money that matters. Suppose that the level of bank lending (i.e., the stock of lending) is constant from one year to the next, and therefore no transactions whatsoever are financed by a bank loan. What happens to the level of payments, assuming that the quantity of money is given, and that the actual turnover velocity of money is at the desired figure?

The answer – the devastating answer – is ‘nothing’. No new lending at all is taking place and the level of bank lending is constant, but money continues to determine payments, spending and so on. Further, if the quantity of money were to double because banks acquired more claims on the state and/or other countries, the equilibrium value of payments – and national income – would also double.   The constancy of bank lending would not make a ha’p’orth of difference.

Can we please return to standard monetary economics

I have in various places attacked creditism as a false and dangerous set of ideas. In a 2009 piece for Standpoint my description of it was emphatic, that it was the ‘particular line of thought’ which had ‘undue prominence in policy-making during the most traumatic period and must carry a large share of the blame for what went wrong’. (The piece re-appeared – with minor amendments – as essay 18 in Money in a Free Society.) I have now shown in this note that no meaningful relationship obtained in the UK’s Great Recession either between the level of bank lending and the level of nominal GDP, or between the change in bank lending and the change in nominal GDP. I have also explained that – when viewed relative to the total value of payments in an economy – the irrelevance of bank lending by itself is obvious. (I accept that bank lending matters enormously, to the extent that it affects the quantity of money.) The proportion of total transactions financed by new bank lending is trivially small. Further, money is used to make payments and settle debts, not credit. (Indeed, the very word ‘credit’ implies that debts have not been settled.) All that is required to enable the vast volume of payments in a modern economy is a particular quantity of money. New bank lending can come to a complete halt, and yet still that volume of payments  can continue to be made. Creditism is an intellectual blunder of the first order.

My main purpose here is to enter a plea for a return to standard monetary economics, that is, the monetary economics in which analysts examine money demand functions, recognise that agents are constantly seeking to equilibrate their money holdings with their underlying money demand, and so on. The regulatory follies of the Great Recession have meant a series of years in which the stock of bank lending to the private sector has shrunk. Apparently, according to the banks, there are a few more similar years still to come. (When will the fools in the Bank of England, FSA etc. wake up?) In this sort of world the quantity of money can grow only if banks acquire more claims on the state. That is easy enough to arrange and, by means of QE, has indeed been arranged in the UK since early 2009. But we still have newspapers, politicians and the rest lambasting banks because they are supposed not to be ‘lending enough to small businesses’ and the like.

Tags: , , , , , , , , ,

Leave a Reply

Your email address will not be published.

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Time limit is exhausted. Please reload CAPTCHA.

News Archive

Article Archive