If I taught macroeconomics, I’d get students to read a lot of operational documents before doing anything else. Let’s be clear on the institutional, operational facts before we start modelling anything.
In the UK, I’d have the class read things like the Bank of England Redbook, HM Treasury’s Debt and Reserves Management Report, the Whole of Governments Accounts, and some of those helpful BoE Quarterly Reports explaining Quantitative Easing and money creation.
It wouldn’t be scintillating reading (then neither is textbook macro), but it would get us off a lot of red herring trails before we started looking at models. Models that ignore important institutional facts could just be discarded.
Here’s an example of the sort of thing you could throw away (from Mankiw’s textbook):
It’s the cursed ‘loanable funds’ model, explaining how government deficits ‘crowd out’ private investment by taking up some of the supply of funds available to private investors. The idea is very simple: when the government borrows savings in order to deficit spend, these savings are no longer available to private investors. A basic supply-demand analysis shows that the price of borrowing – the interest rate – must go up, since you now have less supply for the same demand.
Students on an operationally-based macro course would know that this was nonsense, because they would have read this from the Bank of England’s Quarterly Report:
What this tells you is that the Bank of England sets its policy interest rate and lets the amount of reserves needed for lending float at that rate. There are, in other words, always as many ‘loanable funds’ available for lending as there is demand for loans at the rate set by the BoE (plus whatever banks add on as risk premium, etc.).
Students then won’t have to be confused by the question of why, if government deficits crowd out private investment, austerity is criticised by economists like Paul Krugman. The loanable funds analysis seems to suggest that austerity is a good thing. Reducing deficits makes room for private investment, and private investors (all the textbooks tell us) make better investments than governments, since they’re subject to competition and have to wear their own losses rather than passing them on to the taxpayer.
Krugman’s answer is that the loanable funds model breaks down in a ‘liquidity trap’. I am profoundly disturbed by the number of people who claim to reject neoclassical economics as ideological claptrap and then lavish praise on Krugman and other ‘left-wing’ neoclassicals who propound the ‘liquidity trap’ analysis. The analysis is based on the very same loanable funds model. All that is thrown in is a stipulation that the demand for loanable funds, in times of extreme recession, falls so low that that the market-clearing interest rate is below zero (this is from Krugman’s lesson to Niall Ferguson):
The problem, supposedly, is that the effective interest rate can’t really go negative: you won’t have people paying to lend their money (leaving aside some interesting central bank experiments). So the market for loanable funds is out of equilibrium: at the real interest rate of zero, supply exceeds demand.
In this case it’s supposed to be sensible for the government to use its ‘crowding out’ power, in order to drive the equilibrium interest rate up to where it can match the effective real interest rate. This is just a straightforward application of the loanable funds model found in Mankiw’s textbook. The ideal government stimulus, in Krugman’s view, would have the effect of moving the supply of loanable funds from S to S’:
Thus he writes that ‘government borrowing … gives some of those excess savings somewhere to go’ – in loanable funds terms, excess loanable funds are simply removed from the market, driving the equilibrium interest rate up to zero.
That makes the argument against austerity – indeed for fiscal stimulus. But it’s based on the loanable funds model, and so it flies in the face of operational realities. Moreover, a really hardline austerian will counter by bringing up the matter of ‘Ricardian equivalence’ and arguing that the government can’t actually shift the supply curve of loanable funds, since the curve is meant to represent how many such funds the market wants to supply at a given price. If the government removes savings from the market by running a deficit, why doesn’t the market just get back to the amount it wants to supply by saving more, to replace what the government has taken?
It’s no good trying to reject this argument as special pleading. The supply curve in the loanable funds model is meant to represent the decisions of rational economic agents concerning how much they wish to save, factoring in all their expectations of what will happen within the market in the future. Why can’t such agents also factor in what the government will do in the future? It’s easier to predict what the government will do than to predict what the market will do: the government at least announces its plans and sometimes sticks to them.
Krugman can and does reply that this is simply unrealistic: ‘how many people do you know who decide how much they can afford to spend this year by trying to estimate what current fiscal decisions will mean for their taxes five or ten years in the future?’ (from End This Depression Now!). Indeed, but how realistic is it to suppose that there is a single set of points defining the amount that the market will want to save given every possible interest rate?
These fiddly arguments about the ‘realism’ of a decidedly abstract model are what you get to avoid if you throw out the loanable funds model on first sight, recognising that it doesn’t correspond at all to how lending operations actually work in a modern economy. This is precisely what macroeconomics students would do if they began with the operational realities and then turned to looking at models.
Does this mean that they would be left with no models at all when thinking about macroeconomic policy? Not at all. You just need models that are compatible with operational reality. Krugman himself uses such a model in one of his blog posts. Scott Fullwiler has an excellent post showing how this model provides a much better framework for understanding macroeconomics than what is found in the textbooks.
The model simply plots the desired surplus/deficit positions of two sectors – the government sector and the non-government sector, including the external sector, against GDP.
The basic principles determining the slopes of these two curves are: (1) as GDP increases the non-government sector will want to save more and (2) as GDP increases the government will move from deficit to surplus, due mostly to automatic stabilizer effects and the increased tax revenue that comes with GDP growth. The signs on the curves are the reverse of each other, thus the NGFB+ curve (non government financial balance) has positive values for surplus and negative values for deficit, whereas the GSFB- (government sector financial balance) has positive values for deficit and negative values for surplus. The intersection of the two curves gives you the outcome of government fiscal policy:
In other words, total income is a function of how much the government spends into the economy by running a deficit (or takes out of the economy by running a surplus) and how much of this is saved by the non-government sector. It is a stock-flow consistent model, recognising that any spending exceeding revenue by the government must be held as savings (income exceeding spending) by the non-government sector, since the two sectors form a closed financial system.
What does this model allow us to understand? In the first place, it readily shows how fiscal expansion by the government can increase GDP, if desired savings by the non-government sector remain the same. A fiscal expansion will push up the GSFB+ curve, so that it intersects the NGFB- curve at a point further to the right (from A to B):
So there you have a much better argument against austerity than the flawed ‘liquidity trap’ argument.
Of course it isn’t a knock-down argument. It’s still possible to argue that the positive effect on GDP will be countered by ‘Ricardian’ effects. The private sector might increase its net savings desires (for whatever reason) in response to the fiscal expansion. This, also, can be easily modelled: just have the NGFB+ curve shift upwards in response to the shift in the GSFB- curve. It’s a virtue of the model that it doesn’t force you to say that a given result will arise from a particular policy decision. Rather, it allows various possibilities to be modelled, which can then be tested against various real cases.
The model also allows for the inclusion of the real constraint on government deficit spending, which is inflation. At a certain point, we can say, the NFGB+ curve becomes vertical: people have no more savings desires in a currency they don’t want to hold because it loses value.
One commentator on Fullwiler’s blog complained that the model doesn’t include interest rates. Fullwiler replied that it does, just not on the axes. More precisely, the effects of changes in interest rates can be incorporated into the model. If a rise in interest rates increases net savings desires, this can readily be modelled as a shift of the NFGB+ curve upwards.
Fullwiler also notes that not having interest rates on the axes is an advantage of the model, since it frees us from another constraint of the loanable funds model, which forces us to suppose that changes in the interest rate have a homogenous effect on the level of desired saving. This, of course, need not be true. A lower interest rate reduces the incomes of recipients of interest payments from the government; spending might be lower as a result. A higher interest rate might lead to an increase in forward prices relative to spot prices, leading to increased spending. All these real possibilities can fit into this model; they can’t fit into loanable funds. Loanable funds also suffers from having to construct a single ‘average’ interest rate operating over the economy as a whole, which is an unrealistic and unnecessary stricture.
Neoclassicals like Krugman will no doubt complain that this model doesn’t reveal anything about the long-term sustainability of government debt: the worry that continued borrowing by the government might push interest rates on Treasury debt higher than the long-term rate of GDP growth. Students who studied operational macro would perceive the emptiness in this complaint. They would realise that the rate of interest on Treasury bonds of all maturities is a function of the policy rate set by the central bank plus people’s predictions about that policy rate.
After all, they would have read this, from the HM Treasury Debt Management Office:
Students of operational macro would accordingly realise that the interest rates on various Treasury securities can’t be allowed to diverge wildly from the policy rate set by the Bank of England. They would see that there are institutional frameworks in place to prevent this from happening. They would conclude that rates on Treasury debt can’t be left up to the bond market, lest Treasury debt operations provide an arbitrage opportunity that would ‘conflict with the operational requirements of the Bank of England for monetary policy implementation’.
They would also appreciate the risibility of Krugman’s imagined scenario, in which Treasury bonds are rejected by the market altogether, deficits stay as excess reserve balances, and this drives ‘hyperinflation’. They would have read this, from the Bank of England redbook, explaining that the Bank always has the power to add and drain reserves as necessary to maintain its policy rate:
They would notice that this passage says nothing about the size of the government deficit, since its authors recognise that that is completely irrelevant to the Bank’s power to add and drain reserves as necessary to sustain monetary policy. And they would know, as we’ve already seen, that banks are never constrained by the volume of available reserves but only by the policy rate.
And so they would see it as another virtue of the proposed model that it allows no scope for discussing the long-term sustainability of government debt, since existing operational frameworks make that a complete non-issue.
The proposed model, in short, isn’t based on any incorrect understanding of operational realities. It allows various possible results of policy choices to be thought through clearly. It makes no troubling assumptions of rational expectations. It’s the kind of model that would, I believe, be far more prominent in macroeconomics if we taught the operational realities first and then went on to try to model the possibilities they allow.