Banks, Aggregate Demand, and Aggregate Supply

What is the relation between banks, aggregate demand, and aggregate supply? Do bad banks shift the AD curve or the AS curve? Do bad banks make it harder for fiscal or monetary policy to shift the AD curve?

Some economists argue that you need to fix the banks (and the financial system) to get an economic recovery; fiscal or monetary policy alone won't do it. I wish people who made that argument would be more clear on where they see banks fitting into the AD and AS framework. Keiichiro Kobayashi for example argues that Paul Krugman puts too much emphasis on fiscal policy rather than fixing bad banks; he argues that fixing the banking system is a necessary condition for recovery, and fiscal policy alone won't work.

I can understand how bad banks could affect the AS curve.

Long-run growth comes from the Long Run Aggregate Supply curve moving slowly rightward over time as savings create investment that adds to the stock of capital and increases productivity. A good banking and financial system will encourage savings and investment, make sure the investments are the most productive ones, and make the LRAS curve move rightwards more quickly over time.

Even in the short run a good banking and financial system will be important in re-allocating capital between growing and declining sectors, if there are shifts in relative demand. If people want fewer cars and more restaurant meals, but banks cannot shift loans from car manufacturers to restaurants, the Short Run Aggregate Supply curve may shift left, because the restaurants won't be able to expand to meet demand, and car manufacturers' prices or wages may be sticky downwards.

If you see the financial crisis as causing the recession by shifting the SRAS curve left, then monetary and fiscal policies, which shift the AD curve right, are not the appropriate cure. Even if you see leftward shifts of the SRAS curve as only part of the story, you will see limits on what monetary and fiscal policy can achieve. When expansionary monetary and fiscal policies start to cause excessive inflation, before output and employment have returned fully to normal, you will know that purely AD policies have reached the limit of what can be expected from them.

But that's not how I interpret Keiichiro Kobayashi's argument. He does not say that increasing demand failed in Japan because it merely caused inflation. (And that would be a hard argument to make, empirically.)

I can understand how bad banks could shift the AD curve left. People and firms borrow and lend through banks (in part). Bad banks widen the spread between borrowing and lending rates. The interest rate that matters for the demand for goods is some sort of weighted average between borrowing and lending rates (because some people and firms are borrowers and others are lenders). The interest rate controlled by central banks is closer to the lending rate. So bad banks widen the spread between the central bank's policy rate and the weighted average rate that matters for spending decisions.

But what is the appropriate response to something that shifts the AD curve left? A policy that shifts it right again. And that should include fiscal and/or monetary policy.

If bad banks affect the economy via their effect on AD, and if fiscal and/or monetary policies cannot work, unless and until we fix the banks, the effect of bad banks on the AD curve must be something more complex than just a simple additive effect.

Let me quote Keiichiro Kobayashi at length:

"According to standard economic theory, fiscal stimulus works well when there is nominal price rigidity, but this condition has nothing to do with the current financial crisis. Western economists are claiming that massive fiscal spending will dispel the pessimism that now prevails in the markets, and generate optimistic expectations. That, they imply, will push up asset prices. This is reminiscent of the theory concerning the effectiveness of fiscal expansion as a tool for fighting an economic downturn that was expounded by John Maynard Keynes during the Great Depression in the 1930s.

Modern economics supports the notion that widespread optimistic expectations contribute to economic recovery. But it has yet to be proven that large-scale fiscal expansion creates these expectations (or causes asset prices to rise). In fact, Krugman is only one of many internationally recognised economists asserting that fiscal spending for general demand stimulus such as public works projects – but not for the disposal of bad assets – can alter macroeconomic expectations. It is hard to fathom why they are advancing such a dubious theory when they must be fully aware that this argument has no theoretically robust basis."

That is not how I thought fiscal policy is supposed to work.

According to the Keynesian story of the transmission mechanism, fiscal policy is supposed to increase AD because, well, the government demands more goods. Or people spend part of their tax cuts. They only talk about asset prices if they worry that increasing demand might cause interest rates to rise, asset prices to fall, and this might offset part of the effects.

According to the monetarist story of the fiscal transmission mechanism, fiscal policy works (if it does work) through increasing velocity, by raising interest rates (which reduces asset prices), and so increase MV.

It is monetary policy, not fiscal policy, that is supposed to work, in part, via raising asset prices.

Admittedly, if fiscal policy did succeed in increasing AD, and increasing future output, that would increase firms' profits, and increase some asset prices. But that is more of a side-effect of a successful fiscal policy, albeit one that might reinforce their effectiveness.

If you believe that bad banks make fiscal policy (or an increase in export demand) ineffective, in the sense that fiscal policy cannot shift the AD curve, and so cannot even cause inflation, until the bad banks are fixed, why is that so? Why do bad banks reduce the fiscal multiplier? Does this effect also work in reverse, so that when banks are bad a normally contractionary fiscal policy wouldn't reduce AD either? Or does it only work one way?

I can see the argument that bad banks would make monetary policy less effective in influencing AD. Banks might be scared to lend out an increase in the monetary base, because their capital reserves are insufficient to support increased lending. So the increased monetary base just sits in their currency reserves (or on deposit at the central bank). This ignores other channels in the monetary transmission mechanism, but nevertheless has a point.

Update: Paul Krugman responds to Keiichiro Kobayashi. 

"Kobayashi’s argument, as I understand it, was that in the 90s Japanese firms weren’t able to take advantage of export opportunities because of lack of access to credit. That’s a fairly exotic argument, and I’d like to see some supporting evidence."

Comments

[brendon]

I think we can see the impact of, if not bad banks, broken bank lending channels in Canada through the Bank of Canada's senior loan officers survey. The financial crisis has temporarily severed the connection between monetary policy and credit markets so that monetary easing has had little impact on lending tightness, which as of the first quarter of 2009 was still near record highs. Therefore, fixing the banking system (whatever that means in practice) is a pre-cursor to a recovery. I see fiscal stimulus as a way of bridging the gap until financial markets normalize and monetary policy can again shift the AD curve to the left (as an aside, it is amazing how grad school beats the AS-AD framework out of you, I find myself reaching for my principles of macro textbook more and more these days)

[Andy Harless]

If you believe in the stylized Ec10 Keynesian demand-side model, where consumption depends on income while investment is exogenous and just depends on animal spirits, then Kobayashi's argument doesn't make much sense. But my understanding is that more sophisticated Keynesian demand-side models usually include an accelerator effect, where investment also depends on income. Empirically, investment is a lot more pro-cyclical than consumption, which suggests that investment rather than consumption is what's doing most of the multiplying. If you believe in that kind of model, then I think it's pretty straightforward that anything which limits the ability of investment to respond to income will reduce the effectiveness of fiscal policy.

I think it's empirically plausible, in fact, that the parameters of the investment accelerator and the multiplier make the Keynesian demand-side process unstable, so that, if it weren't for money and prices, a small stimulus could push income up to infinity. At least it's possible to write down a model that has that propoerty, whereas, unless the MPC is >= 1, you can't do it with just the consumption-based multiplier. So, say, if the MPC is 0.5 and the MPI is 0.5, then you get an overall multiplier of infinity, whereas, if a broken banking system reduces the MPI to zero, you get a multiplier of only 2.

[reason]

Fiscal policy is a palliative not a cure. The cure can only come from inflation (as Krugman and Mankiw have both pointed out). Now where is that helicopter gone.

[JKH]

Nick,

Something puzzling here; I thought Krugman has argued for the nuclear option in recapitalizing the banks; i.e. temporary nationalization/reorganization in a very big way. The reason to recapitalize them is to get them lending, where the risk weights of new assets are non-zero, and therefore dependent on capital.

“Banks might be scared to lend out an increase in the monetary base, because their capital reserves are insufficient to support increased lending. So the increased monetary base just sits in their currency reserves (or on deposit at the central bank). This ignores other channels in the monetary transmission mechanism, but nevertheless has a point.”

Half of this is right; half wrong. The point about capital sufficiency with or without reference to the monetary base is the critical point, and what I thought Krugman had acknowledged. But the point about “lending out an increase in the monetary base” is just functionally and technically wrong, as I’ve argued elsewhere until blue in the face. Banks don’t “lend” reserves. Reserves on deposit, which are the pivotal type of bank reserves, are primarily an accounting device for real time mismatches in the flow of funds for individual banks. The fact there's an enormous excess now on deposit on a system wide basis doesn't change this core functionality. The corollary is that banks don’t require current excess reserves in order to lend. And banks as a whole system can’t “get rid” of reserves that the central bank determines by “lending” them. They can convert excess reserves to required reserves, but will only do so if the result makes sense in term of a risk-adjusted return on capital basis. Given that banks are intelligent enough to be aware of a system wide constraint in terms of the Fed’s chosen excess reserve supply, they will not be as foolish as to try and “get rid” of their excess reserve positions individually by unrestrained lending expansion. And this includes buying zero risk treasury debt, particularly in view of interest rate risk. BTW, some people make an eerily similar perception error on the capital side, by identifying bank capital as some sort of asset, as if they don’t know where it is on the balance sheet. Capital is one source of funding for assets, not an asset itself. But these are lesser points in context.

The banking system expands its asset-liability reach on the basis of its capital strength. If the banking system has insufficient capital, it won’t create new assets. And if it doesn’t create new assets, it won’t create new money. Therefore, bank capital is a pretty important private sector catalyst for monetary stimulus in the normal course. (Part of the “deleveraging” effect in total is the specific withdrawal of leverage in the form of less money created by private sector banking system lending, which is consistent with why the Fed has taken over the slack.)

Just an opinion.

P.S. given that bank operating profits are a direct contribution to GDP, where do they fit into AD and AS thinking?

[Patrick]

The sick big banks are withdrawing credit from consumers as they try to reduce risk. Contracting consumer credit is going to hit AD directly. But Meredith Whitney has made the point, and it seems sensible, that there is also a wealth effect; people feel and consequently act poorer when their credit lines are cut, even if they weren't using the credit.

The argument (as I understand it) is based on fears of being liquidity constrained. In the US fear of unemployment and the loss of health care probably amplifies the fear. When people's credit lines get cut, and they have little or no savings, and they are afraid of unemployment, they start to worry about the 'what-ifs'. What if your car breaks down? You dog gets sick? So, as a hedge, people pull back on spending and even start saving. Classic paradox of thrift stuff, no?

[Nick Rowe]

JKH: "P.S. given that bank operating profits are a direct contribution to GDP, where do they fit into AD and AS thinking?"

I am reminded of this quote:
“The Schleswig-Holstein question is so complicated only three men in Europe have ever understood it. One was Prince Albert, who is dead. The second was a German professor who became mad. I am the third, and I have forgotten all about it.” Lord Palmerston.

My colleague T.K. Rymes once tried to explain the banking imputation in GDP to me. One of us gave up. It starts with a paradox: banks have negative value added if you do it the normal way, because they borrow at a high rate and lend at a low rate, so their "output" is less than their "input". You reading this, TK? Wade in if you are. If I remember him correctly, it all explains why Luxembourg has such a ridiculously high GDP (or something). But the rest of us macroeconomists don't even try to fit bank profits into the AD/AS framework. Too small to worry about.

My take on bank reserves:

If an individual bank makes a loan, by crediting someone's deposit account, it stands to lose reserves when the borrower spends that loan, and the cheque is deposited in a different bank. So I stand by my statement that banks lend reserves.

But yes, for the banking system as a whole it is different. If there is no currency drain, and unless the central bank withdraws reserves from the system, the total quantity of all banks' reserves stays the same (though the *ratio* of reserves to deposits will fall).

But I disagree with you, because I insist on sticking to the Nash Equilibrium perspective. Each bank makes its optimal decision taking other banks' decisions as given. They do not choose collectively. The fact that aggregate reserves will (sometimes) stay the same does not enter into each individual bank's choice. If we reject the Nash Equilibrium approach, and assume that players decide collectively, the paradox of thrift disappears, for example, and we could never be in a recession, with deficient aggregate demand, in the first place. Each consumer would understand that aggregate savings cannot diminish if he increases consumption, so all would increase consumption demand. But consumers don't choose this way. Each knows that if he increases demand for consumption, his savings will fall. Players in prisoners' dilemna don't choose this way. Banks don't act like a cartel.

(Maybe in Canada, where there are only 6 banks, the equilibrium = (1/6) cartel. The Cournot-Nash equilibrium approaches monopoly as the number of players goes to 1.) Each bank expects that 1/6th of any loans will return as reserves.)

[Nick Rowe]

JKH: P.S. I have exaggerated my confidence in the empirical correctness of my position on bank reserves because I want a very clear argument between us, by my taking an absolute position, with no nuance.

And yes, it sounds as though Kobayashi did perhaps misunderstand Krugman's position. No matter. If Kobayashi's "Krugman" doesn't exist, we need to invent him.

[Nick Rowe]

brendon: so bad banks shift AD left, and also reduce the effectiveness of monetary policy. Correct? (I agree with you on the role of AD and AS.)

Andy: people have worked out multiplier/accelerator models. They tend to cycle, or hit floors (gross investment cannot be negative) and ceilings (full employment). They were a big thing in the olden days. But those models tended to ignore interest rates and prices. C, I, K, and Y were the only endogenous variables.

I'm still thinking about what you said about bad banks blocking the accelerator part of that mechanism. I think it might be important, but maybe not as obvious as you say. I'm mulling over a potential post.

reason: but in principle, if fiscal policy could increase AD beyond LRAS, and get inflation and expected inflation rolling, it could become self-sustaining, and fiscal stimulus could slowly be withdrawn as expected inflation increases.

Patrick: That sounds plausible as 2 additional channels whereby bad banks shift the AD curve left.

[kevin quinn]

Welcome back, Nick! Here's a different approach which manages to skirt the AS or AD dichotomy. Two firms, each with its own bank. A weakest-link technology: each can invest either 1 or 2; gross return for each equal to 2*the minimum of the two investment levels. The bank gets the entire net return. Each bank chooses to lend either 1 or 2. We have a coordination game where both lend 1 or both lend 2 are equilibria. In the former, each bank lends little because, in effect, others lend little. It just seems to me that something like the following is true - although technological complementarities are probably not the reaon why-: the credit-worthiness of a firm is in part a function of other firms' ability to obtain financing. So we have equilibria where no individual bank wants to lend because no others are lending. And then it's not clear what fiscal policy can do unless it increases animal spirits. (This in not even 1/4 baked, I'm afraid!)

[Scott Sumner]

Nick, I basically agree with you and Krugman. As you know, I think the US and Japanese financial crises are partly a symptom of falling AD. So I think the Japanese guy reverse causation. But I think you were too dismissive of one of his points. I think that some modern new Keynesianism does argue that fiscal stimulus works by changing inflation expectations (if at all.) This is based on Woodford's model. A good example to look at is a paper by one of Woodford's students (Gauti Eggertsson, AER, Sept. 2008) who argues that the recovery of 1933 was caused by expectations that monetary and fiscal stimulus would generate inflation. The other reason to read this paper is that he cites three of my papers. :) I do think that it is hard to argue fiscal stimulus can work without raising inflation expectations.

[Nick Rowe]

Kevin: Thanks!
I think your coordination problem makes sense, but wouldn't it be a problem even if banks were good? And if it's only 2 firms and 2 banks, why couldn't all 4 just meet round a table and come to a deal? This must happen a lot with financing multi-firm projects. All firms, and their banks, must come to an agreement.

I am thinking of some simpler model, where we start with a simple Keynesian Cross, nuke the banks and forbid them to make new loans, then see what happens to the fiscal policy multiplier.

Scott. I probably was too dismissive. He may be onto something (and I suspect he's not alone, and may have empirical support). But it's so unclear what he is arguing. And I don't think it's the expected inflation/real interest rate channel.

I did a blog post on one Gauti Eggertson working paper a few months back. A New Keynesian model, where fiscal policy (tax cuts) did indeed work via expected inflation and real interest rates. (It couldn't work through any other channel, since he had perfect capital markets, infinitely-lived households, and so Ricardian Equivalence.) But the transmission mechanism was very flaky, via taxes on firms' MC curves and price setting and inflation via a Calvo Phillips Curve (which has peculiar properties). (He got a *negative* multiplier for temporary cuts in sales taxes!)

[JKH]

Nick,

(BTW, above I thought I was quoting Kobayashi on reserves.)

But, with regard to “Banks don’t ‘lend’ reserves.” (Moi),

Sounds like you agree roughly at the macro “existential” level, in the sense that the banking system cannot “lend” away reserves from its aggregate position.

But as a general rule, “banks don’t lend reserves” is true at the micro “technical” level as well. Take the US case for example. When bank A lends to customer C, and customer C moves funds to bank B, bank A then settles with bank B by paying fed funds into B’s reserve account. A’s reserve account is debited. B’s is credited. Fed funds are the means of settlement, with a resulting change in reserve distribution. But there is no loan of fed funds from A to C or from A to B. Bank transactions with customers do not provide customers with fed funds. Customers have no access to the fed funds market or bank reserve accounts. Changes in reserve positions between banks are the method of clearing and settling the net impact of bank customer lending, but they are not the substance of the lending itself.

The exception to this is the market among banks to borrow and lend reserves among themselves, which is the fed funds market in the case of the US. An actual loan of reserves from A to B occurs in the fed funds market.

Currency is an odd variation. Banks can hold currency as credit for required reserves. But if they use currency reserves as the mode of payment in customer lending, their reserves are reduced by the same amount. So the bank isn’t “lending reserves” per se, as is the case with interbank fed funds lending. Moreover, banks typically hold vault cash in excess of cash pledged as required reserves, so reserves aren’t always affected in bank to customer currency transactions.

I think what I described isn’t inconsistent with Nash & Co. The idea that individual banks use their knowledge of a collective constraint in optimizing their lending decisions is quite compatible with Nash behaviour. The central bank’s setting for system excess reserves is such a constraint. If a major Canadian bank’s excess reserve position is 1/6 of what the cash manager estimates the system excess to be, that manager may behave differently compared to the case where the position is 2/6 of the estimated total. When a manager in the US system knows that current system excess reserves are 40,000 per cent higher than what they are normally, he’ll take that into account in formulating strategy, including a check on capital position before loading up on additional asset risk. And if he’s got an idea to drive risk-free asset yields down to zero, or negative, in an attempt to transfer his bank's excess reserve position to other banks, he’ll probably consider such aspects as asset availability, asset pricing, expected profit or saving, and potential competitor strategies and counter strategies before proceeding.

(BTW, in the tradition of head hurting ideas, system surplus vault cash is not part of the central bank's excess reserve setting.)