Recent Posts

Sunday, April 15, 2018

Australian Fiscal Surpluses And Functional Finance

Chart: Australian General Government Fiscal Balance (IMF)

The chart above shows the (annual) fiscal balance for the general government sector of Australia (the general government sector includes sub-sovereign governments as well as the central government), taken from the IMF World Economic Outlook. As can be seen, there was a lengthy period of surpluses from the late 1990s - 2000s. This is somewhat unusual, and raises some questions about some interpretations of functional finance and Modern Monetary Theory (MMT). If we look at the situation more carefully, the fact that there were surpluses then was not too surprising, and is entirely consistent with functional finance principles.

(Note: This article was written to respond to a question I received regarding the Australian surpluses. I do not have time to track down Australian economic data, so my description of the conditions during the surplus period is correspondingly vague. And although I watched Australian data as part of the data flow, I never had the need to dig deeply into the Australian situation as part of my own research. The focus here is more on the Functional Finance perspective.)

Australian Surpluses: Not Too Surprising

For the developed countries in the modern era, fiscal deficits are the norm. One can argue that this is the result of basic debt mathematics, which many advocates of balanced budgets conveniently ignore. The usual goal of policy is a "stable" debt-to-GDP ratio. (There is no good theoretical explanation of why this is the case, but that is not the concern here.) Since nominal GDP is growing in most developed countries most of the time (Japan acting somewhat as an exception), governments need to add debt to keep the numerator to grown in line with the denominator. As a result, we do not need a whole lot of theory to predict that the typical outcome is a fiscal deficit.

So the Australian situation is somewhat unusual (but not unprecedented, as discussed later). But there was a lot going on during that period that would allow the general government sector to run surpluses. (Note that I do not have a good feel for the breakdown between the central government and the state governments.)

One of the most important factors was that Australia -- like most of the "Anglo" countries -- has had an impressive bull market in house prices (and construction). The housing bubble in the United States popped, but not in Australia.
Chart: Australian Household Debt-to-GDP

One mechanical effect of higher house prices is that households need to borrow more to purchase them; hence household debt rises. I currently have not imported the full set of the Australian national accounts, but the chart above is indicative of the situation (taken from an IMF report, downloaded using the St. Louis Federal Reserve FRED interface). Unfortunately, it does not overlap the full period of surpluses, but we see that the ratio is rising while greater than 90% of GDP in 2006-2008. The implication is that some households have to emitting enough debt to not only keep up with growing nominal GDP as well raising the ratio. I have not crunched the numbers, but it is clear that the household debt issuance is going to be larger than the general government surplus.

The business sector was presumably growing robustly during this period -- there was first the telecom boom, and then the commodity boom. I would assume that Australian equities also participated in the global equity bull market of that era, and so capital gains taxes would have been elevated. Since capital gains taxes are incurred in a somewhat voluntary fashion, they presumably have less of a dampening effect than income taxes.

Another unusual factor was the privatisation of the national telecoms company, Telstra. Selling off the family silverware was a standard tactic of post-1980 neoliberal politicians, and allows for one-time debt reduction in a relatively painless fashion. (Instead of imposing a tax to reduce the debt level, the government swaps a financial asset for its bonds. A tax reduces the income of the paying entity, while an asset swap is income neutral.)

(Another unusual factor in Australian data is that a significant portion of its current account is intermediated by multinational resource corporations. This means that the international component of the financial balance accounting identities cannot be interpreted as arm's length dealings between Australians and foreigners. The reader would need to find a source more familiar with that aspect of the data.)

The final result is that the surplus was not unusual -- household debt issuance was more than enough to offset the general government financial surplus. Note that this argument is largely theory-free, although it might have problems if one chooses a sufficiently incorrect theory. For example, if one argues that the behaviour of the household sector can be modelled as the behavioural choices of a single representative household, household sector borrowing to purchase housing is neutral for growth, since it represents the same household borrowing from itself.

Bill Mitchell, an Australian MMT academic, summarises the situation in this relatively recent article -- link. He has written detailed analysis of the data on his website earlier (I had not spotted them after a quick search; I will leave the search to an interested reader).

A Blow to Functional Finance?

My questioner wanted to know how this surplus episode related to functional finance (link to primer) and MMT. Was not the argument that fiscal policy is extremely powerful, so how did Australia avoid recession? (Note that Australia has avoided recession for over 20 years.) 

(One possible addition to the confusion is the citation of some MMT authors of a historical oddity in U.S. data -- every time the U.S. federal government has run a surplus, a depression/recession followed relatively soon thereafter. Although this is a cool coincidence that impresses Americans, it is incorrect to say that the MMT line is that fiscal surpluses always cause recessions. Bill Mitchell is one of the lead contributors to MMT and is Australian -- and is well aware of the Australian fiscal surplus.)

From the perspective of functional finance, there is nothing to explain. One of the principles of functional finance is that the level of the fiscal balance does not tell us about the stance of fiscal policy, so the fact that a government is running a surplus or deficit tells us very little. (This will be returned to below.)

However, there is a lot of discussion in the MMT literature that suggests that running surpluses is a bad idea and/or unsustainable in the long term. So long as the economy is growing in nominal terms, the private sector (including external sector actors, like reserve holding foreign central banks) tend to want to hold increasing nominal amounts of financial assets, particularly fixed income. However, there are limits of private sector entities to service debts, so growth driven by private sector debt issuance will generally tend to go off the rails eventually. The Australian housing market sidestepped 2008, but it is hard to see that performance being repeated for decades. Conversely, a central government borrowing in its own free-floating currency faces no risk of involuntary bankruptcy, and so its debts help to stabilise private sector portfolios. However, the inability to sustain surpluses for decades does not imply that surpluses can occur for a few years without a crisis.

Cannot Tell Fiscal Stance By Looking at the Fiscal Balance

I noted the functional finance principle: you cannot tell what the fiscal stance is by looking at the fiscal balance. It is likely that this seems like a vague verbal formulation, but it is a useful theoretical point if we contrast it to conventional thinking. I will now flesh out the hidden details in that formulation.

The writings of functional finance are best understood in the context of existing conventional thinking. One common approach to fiscal policy in DSGE macro is to specify fiscal policy as two exogenous time series:
  1. a time series of government purchases of the composite good in real terms (g);
  2. a time series of lump sum taxes that are fixed in real terms for all time. The "lump sum" specification means that the level of tax is fixed (in real terms) regardless of the wealth or income of the (representative) household; something like a poll tax.
(As a disclaimer, some DSGE models have taxes that are function of economic activity -- distortionary taxes. However, a lot of loose analysis of fiscal policy implicitly is based on lump sum taxes.)

They then make two heroic assumptions (modulo some tweaks, depending on the model):
  1. There is a single composite good in the economy, and so there is only a single good price to worry about, and one notion of production capacity (output gap).
  2. All households are the same: same income, same wealth, pay same taxes.
With only one good to worry about, a tax bill specified in real terms is a near perfect counterpart to government consumption in real terms. Therefore, the two sides of fiscal policy (taxes and spending) are commensurable, and it makes perfect sense to subtract the two time series to see what the net effect on the economy is. In other words, the fiscal balance is a good summary of the action of fiscal policy. (If one actually solved the DSGE macro problem, this would not be strictly correct. However, these models are not actually solved, and the entire action of fiscal policy is ignored, so it does not really matter. This earlier article discusses this rather strange subject.)

Another side effect of this "lump sum tax thinking" is that the fiscal balance is an exogenous variable -- arbitrarily determined by policymakers independent of the state of the economy. Courtesy of Economics 101 mis-education, pretty well every media discussion of fiscal balances implies that governments choose the level of the fiscal deficit each year. (Mainstream academics might complain that this is unfair; the models with distortionary taxes depend on the level of output. However, they were the brain trust that uses the lump sum tax models in Economics 101, so they are part of the problem, not the solution.)

The correct way of thinking about fiscal policy is start off with the parameters that the government actually controls (which could be viewed as exogenous).
  • Government spending might appear straightforward, but it is actually a mix of income transfers, salary spending with wages set in multi-year settlements, contractual purchases of some goods and services, and open market purchases of others. Modern welfare states have programmes that are sensitive to the state of the economy, such as unemployment insurance and welfare. The demand effect is hitting a wide variety of goods and services, each with own capacity constraints. Obviously, if the government ramps up spending on goods that are in short supply, it will exacerbate price increases, whereas purchases where there is excess capacity will have little to no effect on prices. 
  • Tax codes are complicated, and come in a variety of forms. Even the base income tax rate is technically a continuum of tax rates defined over the set of incomes (although we can reduce the complexity by just looking at brackets and marginal tax rates). 
In other words, the government can control tax rates, but it cannot control how much taxes it collects. The fiscal balance is an outcome, not a variable that can directly set by policymakers.

The outcome of the fiscal balance depends upon the nature of the economy: the external balance, investment trends, etc. For a country successfully pursuing an export-led growth strategy -- e.g., Japan up until the 1990s -- the government is likely going to tend towards surplus. Conversely, we should not be surprised that fiscal deficits are high in countries with ageing populations, current account deficits, and anaemic private sector investment.

In summary, the fact that Australia ran a series of surpluses is not an issue for functional finance.

Reduced Form Models -- Yicch

I periodically see complaints that MMT has "no models or theory." From context, the complainers appear to be looking for reduced-form mathematical models. The fact that MMT avoids these reduced-form models is an advantage in my opinion -- despite the fact that my training is as an applied mathematician, and "quantitative analyst" appeared in one form or another on my business card for my career in finance. And one may note that I am hardly alone -- there's a small sub-culture of MMTers in rates finance, and modern rates investors are not normally thought of as literary theorists.

The reason to avoid reduced form models is obvious to anyone with markets experience: pretty much every possible reduced-form mathematical model has been thrown at the macro data set, using every known fitting technique in applied mathematics. The results are uniformly dismal: at best, one can find a model that works for a period of time, then blows sky high when the regime shifts. This is exactly what (good) theory predicts. We can view observed data as being the outcome of a stock-flow consistent model with really complex behavioural functions. We are likely able to approximate those behavioural functions around some "operating point" -- which means that we can fit some reduced form model (with sufficient free parameters) to a limited run of data. Once we move away from the operating point, the model breaks down. Every competent, experienced rates investor knows this, so there is no reason to get excited about reduced-form models.

(DSGE macro side-steps the problem by not actually solving the models, rather just philosophising what the solution might look like -- assuming that the solution exists. From my vantage point, that's boring literary criticism, not mathematics.)
Chart: Velocity of M2 in the United States
Most failed reduced form models are proprietary, and so their track record is not publicly documented. However, we do have a spectacular documented example -- classical Monetarism. The standard interpretation of Milton Friedman's writings was that reduced form models based on (near) constant money velocity were valid. The implication is that all policy makers need to do is keep the money supply growing at k% a year, and we would achieve policy nirvana.

Interest in Monetarism surged, and the theory was put into practice (starting around the 1970s, depending on the central bank). The chart above shows the dismal record of the concept. Velocity was stable -- until policymakers started paying attention to the monetary base. Velocity promptly started trending in both directions, making a mockery of money targeting as policy tool. (Somewhat unsurprisingly, the remaining supporters of Monetarism downplay the reductionist version of money supply targeting.)

Although it is easy to ridicule the models in retrospect (like I just did), the interest in Monetarism did indicate that the simplistic versions of the models appeared to be adequate approximations of reality. Any time one looks at a reduced form model, one needs to ask: why is the track record of the proposed model any more impressive than that of constant velocity models based on pre-1970s data?

Quantifying Functional Finance

We now return to functional finance. One could try to summarise functional finance as: "sufficiently loose fiscal policy will eventually lead to inflation." The problem is quantification, and that is an inherent issue with fiscal policy. We cannot slap the fiscal balance into a reduced form model; we need to look at the variables that are exogenous. The problem is that they resist easy parameterisation.
  • As noted earlier, income tax rates are a continuum, not a single percentage. Giving a tax break to billionaires is going to have less of an impact on growth than a reduction of deductions for public pensions.
  • The modern corporate welfare state (particularly in the United States) is not defined in terms of crass subsidies, rather carve-outs of tax ("tax expenditures"). The entire objective of the exercise is to avoid having a well-defined dollar amount for the subsidy to corporate profits. And these measures are not stable: accelerating depreciation schedules is a standard tactic for fighting economic slowdowns. How do we describe things like changes to depreciation schedules in a model?
  • Spending needs to broken down into its categories, and each acts differently within an economic model.
In order to properly quantify these effects, we would need a relatively complex model. The reality is that such a model will end up with too many free parameters -- and we cannot fit it to data. As a result, any analysis of fiscal policy has to rely on a certain amount of instinct and guesswork. This reflects reality -- nobody can plausibly forecast the detailed effects of the Republican tax cuts on growth.

We are only relatively safe if we imagine hypothetical extreme scenarios. What happens if central government spending is 20% of GDP, yet it cuts all tax rates to zero? Although we might not be able to predict exactly what would happen, it is a safe bet that the price level would rise a lot.

For less extreme policies, the precise outcome depends on the interaction of policy with private sector behaviour, as well as the external sector. One can apply rules of thumb, but we cannot take any implied forecasts too seriously. One can hope that nominal currency amounts reflect the relative importance of spending and taxes, but we have to accept that this is a simplification. (I certainly appeal to this simplification in my own writings on occasion.)

(c) Brian Romanchuk 2018

No comments:

Post a Comment

Note: Posts are manually moderated, with a varying delay. Some disappear.

The comment section here is largely dead. My Substack or Twitter are better places to have a conversation.

Given that this is largely a backup way to reach me, I am going to reject posts that annoy me. Please post lengthy essays elsewhere.