MMT

Kelton and Krugman on IS-LM and MMT

The MMT debates continue apace. New critiques — the good, the bad and the ugly — appear daily. Amidst the chaos, a guest post on Alphaville from three MMT authors stood out: the piece responded directly to various criticisms while discussing the policy challenges associated with controlling demand and inflation when fiscal policy is the primary macro stabilisation tool. These are the debates we should be having.

Unfortunately, it is one step forward, two steps backwards: elsewhere Stephanie Kelton and Paul Krugman have been debating across the pages of the Bloomberg and the New York Times websites. The debate is, to put it politely, a mess.

Krugman opened proceedings with a critique of Abba Lerner’s Functional Finance: the doctrine that fiscal policy should be judged by its macroeconomic outcomes, not on whether the financing is “sound”. Lerner argued that fiscal policy should be set at a position consistent with full employment, while interest rates should be set at a rate that ensures “the most desirable level of investment”. Krugman correctly notes the lack of  precision in Lerner’s statement on interest rates. He then argues that, “Lerner neglected the tradeoff between monetary and fiscal policy”, and that if the rate of interest on government debt exceeds the rate of growth, either the debt to GDP ratio spirals out of control or the government is forced to tighten fiscal policy.

Kelton hit back, arguing that Krugman’s concerns are misplaced because interest rates are a policy variable: the central bank can set them at whatever level it likes. Kelton points out that Krugman is assuming a “crowding out” effect: higher deficits lead to higher interest rates. Kelton argues that instead of “crowding out”, Lerner was concerned about “crowding in”: the “danger” that government deficits would push down the rate of interest, stimulating too much investment. Putting aside whether this is an accurate description of Lerner’s view, MMT does diverge from Lerner on this issue: since MMT rejects a clear link between interest rates and investment,  the MMT proposal is simply to set interest rates at a low level, or even zero, and leave them there.

So far, this looks like a straightforward disagreement about the relationship between government deficits and interest rates: Krugman says deficits cause higher interest rates, Kelton says they cause lower interest rates (although she also says interest rates are a policy variable — this apparent tension in Kelton’s position is resolved later on)

Krugman responded. This is where the debate starts to get messy. Krugman takes issue with the claim that the deficit should be set at the level consistent with full employment. He argues that at different rates of interest there will be different levels of private sector spending, implying that the fiscal position consistent with full employment varies with the rate of interest. As a result, the rate of interest isn’t a pure policy variable: there is a tradeoff between monetary and fiscal policy: with a larger deficit, interest rates must be higher, “crowding out” private investment spending.

Krugman’s argument involves two assumptions: 1) there exists a direct causal relationship between the rate of interest and the level of private investment expenditure, and, 2) the central bank will react to employment above “full employment” with higher interest rates. He illustrates this using an IS curve and a vertical “full employment” line (see below). He declares that “this all seems clear to me, and hard to argue with”.

250219krugman1-jumbo

At this point the debate still appears to remain focused on the core question: do government deficits raise or lower the rate of interest? By now, Krugman is baffled with Kelton’s responses:

It seems as if she’s saying that deficits necessarily lead to an increase in the monetary base, that expansionary fiscal policy is automatically expansionary monetary policy. But that is so obviously untrue – think of the loose fiscal/tight money combination in the 1980s – that I hope she means something different. Yet I can’t figure out what that different thing might be.

This highlights two issues: first, how little of MMT Krugman has bothered to absorb, and, second, how little MMTers appear to care about engaging others in a clear debate. Kelton, following the MMT line, is tacitly assuming that all deficits are monetised and that issuing bonds is an additional, and possibly unnecessary, “sterilisation” operation. Under these assumptions, deficits will automatically lead to an increase in central bank reserves and therefore to a fall in the money market rate of interest. But Kelton at no point makes these assumptions explicit. To most people, a government deficit implicitly means bond issuance, in correspondence with the historical facts.

So Krugman and Kelton have two differences in assumptions that matter here. First, Krugman assumes a mechanical relationship between interest rates and investment and thus a downward sloping IS curve, while Kelton rejects this relationship. Second, they are assuming different central bank behaviour. Krugman assumes that the central bank will react to fiscal expansion with tighter monetary policy in the form of higher interest rates: the central bank won’t allow employment to exceed the “full employment” level. Kelton assumes, firstly, that fiscal policy can be set at the “full employment” level, without any direct implications for interest rates and, secondly, that deficits are monetised so that money market rates fall as the deficit expands.

The “debate” heads downhill from here. Krugman asks several direct questions, including “[does] expansionary fiscal policy actually reduce interest rates?”. Kelton responds, “Answer: Yes. Pumping money into the economy increases bank reserves and reduces banks’ bids for federal funds. Any banker will tell you this.” Even now,  neither party seems to have identified the difference in assumptions about central bank behaviour.

The debate then shifts to IS-LM. Krugman asks if Kelton accepts the overall framework of discussion — the one he previously noted “all seems clear to me, and hard to argue with”. Kelton responds that, no, MMT rejects IS-LM because it is “not stock-flow consistent”, while also correctly noting that Krugman simply assumes that investment is a mechanical function of the rate of interest.

In fact, Krugman isn’t even using an IS-LM model — he has no LM curve — so the “not stock flow consistent” response is off target. The stock-flow issue in IS-LM derives from the fact that the model solves for an equilibrium between equations for the stock of money (LM), and investment and saving (IS) which are flow variables. But without the LM curve it is a pure flow model: Krugman is assuming, as does Kelton, that the central bank sets the rate of interest directly. So Kelton’s claim that “his model assumes a fixed money supply, which paves the way for the crowding-out effect!” is incorrect.

Similarly, Kelton’s earlier statement that Krugman “subscribes to the idea that monetary policy should target an invisible ‘neutral rate'” makes little sense in the context of Krugman’s IS model: there is no “invisible” r* in a simple IS model of the type Krugman is using: the full employment rate of interest can be read straight off the diagram for any given fiscal position.

Krugman then took to Twitter, calling Kelton’s response “a mess”, while still apparently failing to spot that they are talking at cross purposes. Kelton hit back again arguing that,

The crude, IS-LM interpretation of Keynes demonstrates that, under normal conditions, an increase in deficit spending will push up interest rates and lead to some crowding-out of investment spending. There is no room for a technical analysis of monetary operations in that framework.

Can this discussion be rescued? Can MMT and IS-LM be reconciled? The answer, I think, turns out to be, “yes, sort of”.

I wasn’t the only person pondering this question: several people on Twitter went back to this post by Nick Rowe where he tries to “reverse engineer” MMT using the IS-LM model, and comes up with the following diagram:

Rowe-IS-LM
Does this help? I think it does. In fact, this is exactly the diagram used by Victoria Chick in 1973, in The Theory of Monetary Policy, to describe what she calls the “extreme Keynesian model” (bottom right):

Chick-Theory-Monetary-Policy-scaled

So how do we use this diagram to resolve the Krugman-Kelton debate? Before answering, it should be noted that MMTers are correct to point out problems with the IS-LM framework. Some are listed in this article by Mario Seccareccia and Marc Lavoie who conclude that IS-LM should be rejected, but “if one were to hold one’s nose,” the “least worst” configuration is what Chick calls the “extreme Keynesian” version.

To see how we resolve the debate, and at the risk of repeating myself, recall that Krugman and Kelton are talking about two different central bank reactions. In Krugman’s IS model, the central bank reacts to looser fiscal policy with higher interest rates. Kelton, on the other hand, is talking about how deficit monetisation lowers the overnight money market rate. Kelton’s claim that a government deficit reduces “interest rates” is largely meaningless: it is just a truism. Flooding the overnight markets with liquidity will quickly push the rate of interest to zero, or whatever rate of interest the central bank pays on reserve balances. It is a central bank policy choice: the opposite of the one assumed by Krugman.

But what effect will this have on the interest rates which really matter for investment and debt sustainability: the rates on corporate and government debt? The answer is “it depends” — there are far too many factors involved to posit a direct mechanical relationship.

This brings a problem that is lurking in the background into sight. Both Kelton and Krugman are talking about “interest rates” or “the interest rate” as if there were a single rate of interest, or that all rates move together — the yield curve shifts bodily with movements of the policy rate. As the chart below shows, even for government debt alone this is a problematic assumption.

yc

Now, in the original IS-LM model, the LM curve is supposed to show how changes in the government controlled “money supply” affects the long term bond rate of interest. This is because, for Keynes, the rate of interest is the price of liquidity: by giving up liquidity (money) in favour of bonds, investors are rewarded with interest payments. But the problem with this is that we know that central banks don’t set the “money supply”: they set a rate of interest. So, it has become customary to draw a horizontal MP curve, allegedly representing an elastic supply of money at the rate of interest set by the central bank. But note that in switching from a sloping to a horizontal LM curve, the “interest rate” has switched from the long bond rate to the rate set by the central bank.

So how is the long bond rate determined in the horizontal MP model? The answer is it isn’t. As in the more contemporary three-equation IS-AS-MP formulation, it is just assumed that the central bank fixes the rate of interest that determines total spending. In switching from the upward sloping LM curve to a horizontal MP curve, the crude approximation to the yield curve in the older model is eliminated.

What of the IS curve? Kelton is right that a mechanical relationship between interest rates and investment (and saving) behaviour is highly dubious. If we assume that demand is completely interest-inelastic, then we arrive at the “extreme Keynesian” vertical IS curve. But does Kelton really think that sharp Fed rate hikes will have no effect on total spending? I doubt it. As Seccareccia and Lavoie note, once the effects of interest rates on the housing market are included, a sloping-but-steep IS curve seems plausible.

Now, does the “extreme Keynesian” IS-LM model, all the heroic assumptions notwithstanding, represent the MMT assumptions? I think, very crudely, it does. The government can set fiscal policy wherever it likes, both irrespective of interest rates and without affecting interest rates: the IS curve can be placed anywhere along the horizontal axis. Likewise, the central bank can set interest rates to anything it likes, again without having any effect on total expenditure. This seems a reasonable, if highly simplified approximation to the standard MMT assumptions that fiscal policy and monetary policy can be set entirely independently of each other.

But is it useful? Not really, other than perhaps in showing the limitations of IS-LM. The only real takeaway is that we deserve a better quality of economic debate. People with the visibility and status of Kelton and Krugman should be able to identify the assumptions driving their opponent’s conclusions and hold a meaningful debate about whether these assumptions hold — without requiring some blogger to pick up the pieces.

Misunderstanding MMT

MMT continues to generate debate. Recent contributions include Jonathan Portes’ critique in Prospect and Stephanie Kelton’s Bloomberg op-ed downplaying the AOC and Warren tax proposals.

Something that caught my eye in Jonathans’ discussion was this quote from Richard Murphy: “A government with a balanced budget necessarily denies an economy the funds it needs to function.” This is an odd claim, and not something that follows from MMT.

Richard has responded to Jonathan’s article, predictably enough with straw man accusations, and declaring, somewhat grandiosely, that “the left and Labour really do need to adopt the core ideas of modern monetary theory … This debate is now at the heart of what it is to be on the left”

Richard included a six-point definition of what he regards to be the core propositions of MMT. Paraphrasing in some cases, these are:

  1. All money is created by the state or other banks acting under state licence
  2. Money only has value because the government promises to back it …
  3. … because taxes must be paid in government-issued money
  4. Therefore government spending comes before taxation
  5. Government deficits are necessary and good because without them the means to make settlement would not exist in our economy
  6. This liberates us to think entirely afresh about fiscal policy

Of these, I’d say the first is true, with some caveats, the second and third are partially true, and the fourth is sort of true but also not particularly interesting. I’ll leave further elaboration for another time, because I want to focus on point five, which is almost a restatement of the quote in Jonathan’s Prospect piece.

This claim is neither correct nor part of MMT. I don’t believe that any of the core MMT scholars would argue that deficits are required to ensure that there is sufficient money in circulation. (Since Richard uses the term “funds” in the first quote and “means [of] settlement” in the second, I’m going to assume he means money).

To see why, consider what makes up “money” in a modern monetary system. Bank deposits are the bulk of the money we use. These are issued by private banks when they make loans. Bank notes, issued by the Bank of England make up a much smaller proportion of the money in the hands of the public. Finally, there are the balances that private banks hold at the Bank of England, called reserves.

What is the relationship between these types of money and the government surplus or deficit? The figure below shows how both deposits and reserves have changed over time, alongside the deficit.

uk-money-supply-deficit

Can you spot a connection between the deficit and either of the two money measures? No, that’s because there isn’t one — and there is no reason to expect one.

Reserves increase when the Bank of England lends to commercial banks or purchases assets from the private sector. Deposits increase when commercial banks lend to households or firms. Until 2008, the Bank of England’s inflation targeting framework meant it aimed to keep the amount of reserves in the system low — it ran a tight balance sheet. Following the crisis, QE was introduced and the Bank rapidly increased reserves by purchasing government debt from private financial institutions. Over this period, and despite the increase in reserves, the ratio of deposits to GDP remained pretty stable.

The quantity of neither reserves nor deposits have any direct relationship with the government deficit. This is because the deficit is financed using bonds. For every £1bn of reserves and deposits created when the government spends in excess of taxation, £1bn of reserves and deposits are withdrawn when the Treasury sells bonds to finance that deficit.

This is exactly what MMT says will happen (although MMT also argues that these bond sales may not always be necessary). So MMT nowhere makes the claim that deficits are required to ensure that the system has enough money to function.

It is true that the smooth operation of the banking and financial system relies on well-functioning markets in government bonds. During the Clinton Presidency there were concerns that budget surpluses might lead the government to pay back all debt, thus leaving the financial system high and dry.

But the UK is not in any danger of running out of government debt. Government surpluses or deficits thus have no bearing on the ability of the monetary system to function.

The macroeconomic reason for running a deficit is straightforward and has nothing to do with money. The government should run a deficit when the desired saving of the private sector exceeds the sum of private investment expenditure and the surplus with the rest of the world. This is not an insight of MMT: it was stated by Kalecki and Keynes in the 1930s.

If a debate about MMT really is at “the heart of what it is to be on the left” then Richard might want to take a break to get up to speed on MMT (and monetary economics) before that debate continues.

MMT meets Rey’s dilemma: a balance sheet view of capital flight (coming soon to an EM country near you)

Recently, a colleague emailed with the following set of questions: ‘a balance sheet approach to defending currencies. Do you know literature that explains in detail the globally interlocking balance sheets between central banks, commercial banks and what happens when a national government has to defend its currency? What is the role of national and foreign reserves and how do they travel these balance sheets in the process of trying to defend a currency? I came back to this question when discussing the Swedish fight to defend the Dollar-pegged Krona in the early 90s and the promise of MMT? Most particularly we wondered to what extent national governments can just issue Krona and use them to buy foreign reserves or what sets the limits exactly to this attempt?’

My MMT friends do have answers to these questions (and they do spend a lot of time defending MMT from critiques that it doesnt consider balance of payment constraints to monetary sovereignty). I thought I would answer these questions a la Minsky, with balance sheets, since that’s how I teach my undegrad students about exchange rate management in emerging/developing countries. I teach by setting those questions within the broader conversations about global liquidity, global financial cycles and Rey’s dilemma – independent monetary policy is only possible if countries manage capital flows (capital controls).

  1. Start with an economy in autarchy: central bank issues reserves to banking sector for settlement purposes (banks pay each other in reserves), banks lend, create bank deposits in the process.

Screen Shot 2018-02-14 at 15.45.46

2a. Commercial bank borrows abroad from parent bank/interbank market (USD/EUR/JPY)

Screen Shot 2018-02-14 at 15.55.58

(this scenario played out in Eastern Europe before Lehman, when foreign-owned banks would borrow from parent/interbank markets – ending up with the Vienna Initiative)

2.b Commercial bank funding via fx swap with non-residents

Screen Shot 2018-02-14 at 15.55.16

Step 1 occurs where local banks are allowed to lend retail in foreign currency. If it looks like MMT 2.0, it is not exactly that – without legal restrictions, the only constraint on banks creating foreign money (eurodollars) is their foreign currency reserves (an exogenous money story a la monetarism).

Even with restrictions on the lending in foreign currency (skipping steps 1&2), banks typically intermediate non-resident demand for local currency bonds via fx swaps (see my paper here on growing appetite for EM securities as part of shadow banking reform agenda). This is big enough that BIS has recently proposed to approach fx swaps as missing debt. Note that this is a global liquidity story:  without capital controls, non-resident demand/bank borrowing abroad reflects funding conditions in US money markets (see Bruno and Shin’s risk taking channel of monetary policy).

3. Rey’s dilemma kicks in: central bank intervenes to stem currency appreciation (for mercantilist or macroprudential reasons)

Screen Shot 2018-02-14 at 15.05.09

For this commercial bank, the central bank’s policy rate is no longer a binding constraint, since it obtains local base money (reserves) by selling its fx liquidity to the central bank, rather than in the local interbank money market. When interest rate differentials are significant, this eases cost of funding (in the macro literature, this is part of the debate on the effect of financial globalisation on the effectiveness of inflation targeting central banks).  It’s global liquidity, not domestic liquidity, that determines short-term money market rates.

4. To regain monetary control, central banks issue own debt.

Screen Shot 2018-02-14 at 15.14.11

This operation is known as sterilisation: that is, ‘sterilising’ the impact of fx market interventions on domestic money market rates. Central bank issues own securities (or sells government bonds, or takes deposits) in order to absorb back the reserves it created when it paid for the fx liquidity it bought from banks. Note here that this does not solve Rey’s dilemma, since banks have full discretion over how much to place in central bank securities. Rather, for banks this is an attractive carry – borrowing cheap abroad, placing it in risk-free local securities (banks can hedge fx risk).

If you think this is a theoretical exercise, think again.

Screen Shot 2018-02-14 at 15.20.55

5. The limits to monetary sovereignty: global liquidity conditions tighten, capital flight ensues.

Screen Shot 2018-02-14 at 15.29.04

In step 1, non-residents sell local securities – potentially triggering liquidity spirals if large, unregulated local repo market exists.  Note that by step 5, local banks with no direct links to global finance also start to suffer as interbank liquidity tightens. Cant the central bank mitigate this by reverse sterilisations, that is, by again insulating fx market interventions from domestic money market dynamics? The lessons from the 1997 Asian crisis, according to the IMF, is to segment domestic money markets, that is, to prevent local banks from lending to (non-resident) speculators:

Because a speculative attack requires the establishment of a net short position in the domestic currency, countries have employed a number of tactics to raise the costs of short positions. When sterilized intervention fails to stem capital outflows, short-term interest rates are allowed to rise, tightening conditions in financial markets and making it more costly for speculators to obtain a net short position by borrowing domestic currency. Frequently, however, an increase in short-term money market rates is transmitted quickly to the rest of the economy; it may therefore be difficult to sustain for an extended period, especially if there are weaknesses in either the financial system or the nonfinancial sector. When high short-term interest rates impose an unacceptable burden on domestic residents, countries may “split” the markets for domestic currency by requesting that domestic financial institutions not lend to speculators. Foreign exchange transactions associated with trade flows, foreign direct investment, and equity investments are usually excluded from such restrictions. In essence, a two-tier system is created that prevents speculators from getting domestic credit while allowing nonspeculative domestic credit demand to be satisfied at normal market rates. (IMF 1997)

Even if the central bank successfully protects local banks  from cross-border volatility triggered by global financial cycles, it can only defend the currency to the extent that it has foreign reserves. It will most likely not wait until it runs out. In the happy scenario, it draws on its swap lines to weather capital flight – but few central banks have that luxury (and ask yourself, how many will actually have it when Donald Trump needs to be consulted on this). The worst case scenario:  IMF/Troika/whoever will lend  – with heavy conditionality.