Moving on from the household budget analogy: part 1

Both sides of Australian politics are wedded to an analogy which relates fiscal policy to everyday household budgeting. In this world, “responsible economic management” means the consistent delivery of budget surpluses — or at least “balancing the budget over the business cycle.” Deficit spending is the government “drawing on the nation’s credit card,” and an ever-increasing public deficit is a “burden” to be borne by our grandchildren. Therefore, the government should “tighten its belt” and “live within its means” …

These claims are repeated by mainstream media outlets as though they are supported by reliable evidence. In fact, this view of the economy contradicts empirical evidence and is not shared even by the majority of economists.

In modern monetary theory, there are two fundamental differences between government and household budgets.

  • Governments don’t need to earn before they spend. A household which spends more than it earns will face bankruptcy once its savings and credit limit are exhausted. In contrast, government spending creates currency, and the government can run budget deficits indefinitely — especially when the economy is growing.
  • Governments cannot accumulate savings. Households usually invest their unspent income in the financial sector, because they need to accumulate wealth before they can spend it. When the government taxes more than it spends, the surplus is removed from the economy — unless the government spends it on infrastructure or a sovereign wealth fund. But governments don’t need surpluses to finance investment.

This post is the first in a series. To start with, I’ll describe a model for thinking about macroeconomics that replaces the household budget analogy. In later posts, I’ll use that model and ideas from modern monetary theory to justify the two claims above.

But first, a disclaimer: this discussion applies only to governments, like those in the the US, the UK and Australia, which are monetarily sovereign. Things are different when the government’s ability to spend its own currency is limited by a gold standard, fixed exchange rate or currency union. I discussed this previously in What caused the Eurozone debt crisis?

The pie model: an intuitive way to think about macroeconomics

At the level of the individual household and business, ‘responsible economic management’ means accumulating money, or nominal wealth. Nominal wealth is closely related to real wealth — the subjective value of real land, goods, labor, buildings, machines and experiences. Macroeconomics has a lot to say about how that relationship can change, but at the household level it’s relatively stable (most of the time). Households accumulate nominal wealth by spending less than 100% of their nominal income.

A change of perspective is needed to understand things at the macroeconomic level, because unlike households, governments can create nominal wealth. We need a new definition for ‘responsible economic management.’

Responsible governments should try to maximise the growth and stability of real wealth by controlling the quantity and distribution of nominal wealth.

You can think about the effect of government policy (and private sector behaviour) by using the mental image of a divisible pie that changes in size. The size of the pie represents national (or global, or regional) real wealth. The segments represent units of nominal currency (dollars, or billions or trillions of them circulating in the economy).

This $40b economy experienced real GDP growth of $8b (20%) and inflation of 5%, becoming a $50 billion economy. (Each segment represents $1b.)

Households always want to minimise spending, because it removes segments from their pie. Government spending is different: it increases the total number of segments in the pie (taxation removes segments). The link between the number of segments and the size of the pie is very complicated. A given level of government spending could increase real wealth in some circumstances and decrease it in others.

This is not to say that the government can or should control the economy. The private sector’s behaviour is probably even more important than the government’s in determining the size of the pie and distribution of the segments — and in endogenous money models the private sector can change the number of segments (money supply) too. We focus on the government not because it should control the economy, but because we can control the government.

All government action (and inaction) has some effect on the distribution of wealth. For example, deficit spending increases the number of segments in the pie. Economists who support government stimulus in recession conditions argue that because productive capacity is going unused, increasing the number of segments will increase the size of the pie.

On the other hand, if the behaviour of the private sector is already enlarging the pie as quickly as possible, adding more segments would increase inflation. Everyone has more money, but life isn’t actually any better. If the private sector was allocating spending more productively than the government, this would reduce the rate of growth of the pie, too.

When the government taxes more than it spends (fiscal surplus), some segments are removed from the pie. Since there are fewer dollars chasing after the same quantity of real wealth, we’d expect deflation. And if the private sector allocates spending more productively than the government, the pie should grow even faster.

Even if the budget is always strictly balanced, by deciding who is taxed and who receives the benefits of spending, the government alters the distribution of wealth. And because of feedback loops and changing trends in private sector behaviour, the distribution and growth of real wealth seems to be unstable anyway, meaning that by remaining neutral, the government endorses the changes it allows to go unmitigated.

We can think about the effect of government policy and private sector behaviour by adding ‘layers’ to the pie model which represent the distribution of wealth when the economy is divided into different sectors. We can divide the economy however we like: by industry, by income quintile, or even by gender. We can even define ‘the economy’ differently: usually as real national GDP, but the model applies equally to global GDP, the stock of national wealth, or the share of wages. The important thing is that we don’t mix up stocks and flows, that segments add up to 100%, and that we can find reliable data that measures the quantities we’re interested in. (Mathematically, we must define a partition of the ‘economy’ we have chosen.)

Layers showing the size of Australia's government in 2010-11 (in red, left) and Australia's income distribution by quintile in 2009-10 (right).

The government can use fiscal and monetary policy to change the number of slices the pie is divided into (mainly with monetary policy) and the number of slices controlled by different groups of people or industries (mainly using fiscal policy). Of course, what we really want the government to do is make the pie bigger, but both:

  • the way that government policy affects the number and distribution of slices (nominal GDP and income distribution); and
  • the way the number and distribution of slices affects the rate of growth of the pie (real GDP growth)

are poorly understood by governments and economists. All economic models are incomplete, and in most cases, they are also inconsistent both internally and with empirical data. That is okay, because economics is a complex science in its infancy — but politicians and economists frequently get away with using incomplete and inconsistent arguments to justify controversial policy decisions.

In the next post, I’ll look at why governments don’t need to earn before they spend, and consider when fiscal policy decisions like large budget deficits become unsustainable. If you’d like to jump ahead, you might want to read Bill Mitchell’s introduction to modern monetary theory using the business card analogy at Barnaby, better to walk before we run.

Data sources

  • Size of Australia’s government as proportion of GDP in 2010-11: A perspective on trends in Australian Government spending, Australian Treasury (pdf at p 29).
  • Distribution of Australia’s income by quintile in 2009-10: Household Income and Income Distribution, Australian Bureau of Statistics, 2011 (pdf at p16).

[Don’t] Give austerity a chance

Stephen Kirchner and Robert Carling plead us to give austerity a chance, citing Alesina and Ardgana’s work on fiscal adjustment to show that austerity is as likely as fiscal stimulus to be followed by a period of expansion – a finding which is allegedly incompatible with the so-called Keynesian attitude that austerity is “a formula for a self-reinforcing economic downturn”. Though the empirical portion of Alesina’s work has been relatively solid (though not entirely without issue) the conclusions drawn and subsequently Kirchner and Carling’s interpretation of them range between disingenuous, biased and plain wrong.

In the introduction to their recent paper on fiscal adjustments, Alesina and Ardgana demonstrate from the outset that they don’t understand the modern monetary system:

If agents believe that the stabilization is credible and avoids a default on government debt, they can ask for a lower premium on government bonds. Private demand components sensitive to the real interest rate can increase if the reduction in the interest rate paid on government bonds leads to a reduction in the real interest rate

The authors presuppose here that the interest rate paid on government bonds is reduced by a reduction in the amount of government debt. MMT provides a model for understanding why issuing government debt should in fact quickly cause the interest rate to fall – which, by the way, it does:

The blue line represents the current budget deficit as a percentage of GDP. The red line represents the real interest rate paid on government debt. Observe that the short-term relationship is generally inverse – when the government ramps up spending, the fed funds rate falls almost simultaneously. This is consistent with the MMT assertion that government spending creates more bank reserves to compete for government bonds, pushing the interest rate down in the short term1. It is inconsistent with the idea proposed above that “credible stabilisation” (a fancy term for deficit reduction) will enable market participants to “ask for a lower premium on government bonds” – the reaction is just the opposite.

The paper cites an earlier study by the same authors which examines every period in each of the “major OECD nations” since roughly 1980 of fiscal “adjustment” (deficit reduction) and “stimulus” (running higher deficits, sometimes confusingly referred to as ‘fiscal expansion’) to determine whether they were “successful” (sustained for at least three years) and/or “expansionary” (followed by a period of GDP growth outdoing 75% of the other nations in the study). In their own words, “they define a period of fiscal adjustment as a year in which the cyclically adjusted primary balance improves by at least 1.5 per cent of GDP.”

The “cyclically adjusted primary balance” refers to the yearly government budget deficit (or surplus) before counting interest payments (or receipts), which is then adjusted to negate any changes to unemployment. The purpose of this adjustment is to account for changes in welfare costs and tax revenue that might falsely appear as periods of consolidation or stimulus. It is to avoid selecting periods in which discretionary government spending stays the same but exogenous changes to income or expenditure change the deficit-to-GDP ratio. Alarm bells should be ringing at the idea of abstracting out the effects of fiscal policy on unemployment, but the intent is clear enough. Though a rudimentary and non-standard adjustment for this branch of economics, the authors insist that it is unimportant. They conclude their explanation of the cyclical correction process by saying that “even not correcting at all would give similar results.”

Issues with the mathematical determination of fiscal adjustment notwithstanding, the results seem to be at extreme odds with the conclusion that the authors draw from them. The report finds that “fiscal adjustments on the spending side are almost as likely to be associated with high growth (i.e. a successful episode) than fiscal expansions on the spending side” [emphasis added]. In English, this means that increases to government spending are more likely to be successful at promoting economic growth than spending cuts in the general case.

Though it weakens the argument being presented for austerity, it is not much of a vindication in itself for fiscal stimulus. One reason is that the margin is admittedly close, but more importantly no one is seriously arguing that fiscal stimulus is always the best option for growth. By framing the problem in the broad context of all periods of fiscal adjustment, much of the nuance of the Keynesian argument for fiscal stimulus is lost. Keynes made no general statement about the effect of fiscal adjustment on economic growth; in fact he expended considerable effort detailing the problems with studying parts of the economy in isolation.2 A study in the context of the fundamental problems that stimulus is intended to solve is required to construct a compelling argument for it. The fundamentals for demand-side stimulus include unemployment and the private debt level.

In similar spirit to the deferral of interest rate determination to the exogenous forces of “expectations”, Alesina and Ardgana explain that “politics” and ultimately the expectations of voters are responsible for the fact that 85% of the attempted fiscal adjustments in their study were “unsuccessful” and quickly reversed. The study found just 17 periods of successful fiscal adjustment from a sample of 107 attempts. Far from being a purely political phenomenon, the economics of this trend are simple: if deficit reduction activities damage aggregate demand enough that unemployment rises, the government has simultaneously lost sources of tax revenue and gained new welfare recipients – which has further knock-on effects to aggregate demand. Despite attempting to “cyclically adjust” away the effect of unemployment in their figures the point is not completely lost on the authors, who note with a parenthetical lack of surprise that “the spending cuts which have led to sharper and more permanent debt/GDP ratio reductions are those which have stopped the growth of entitlements”.

None of these mistakes are as egregious or insulting than Kirchner and Carling’s Hoover-esque request that we as voters and commentators should be patient and “give austerity a chance”. It is all too easy to ask those suffering to be patient when you are not among them. For the millions of jobless struggling to make ends meet on inadequate welfare assistance, for the homeowners months behind on mortgage payments facing foreclosure, for the small business owners unable to find enough revenue to continue operating, austerity means one thing: more suffering. The situation can only improve when the people have enough financial security to resume their normal lives. Jobs will only be created after people are spending enough to provide business with the revenue to hire more workers. To a private sector already leveraged to the hilt with debt and increasingly unable to make the payments, the only path to recovery is deleveraging. Austerity can only make this painful process more difficult, a budget surplus makes it virtually impossible. Austerity has had plenty of chances. Let’s try something else.

[1] In the long run, since fiscal expansion may be inflationary, the central bank may respond by raising interest rates. This is the weak correlation between the plotted lines visible over a period of decades.

[2] The concept of the ‘fiscal multiplier’ – sometimes called a Keynesian multiplier – actually has very little to do with Keynes at all. It was first proposed by Richard Kahn in a 1931 publication, and was crystallised in John Hick’s IS-LM model, which was introduced as a mathematical model of Keynes’ central ideas to his General Theory. Hicks’ later admitted that IS-LM was merely one of his own older models, rewritten in Keynes’ unusual (for the time) terminology, but it continues to be representative of “Keynesian economics” today.

Extended trading hours and the retail labour market

This was a rather rushed article for a local newspaper. I thought some readers here might be interested too.

In the public debate surrounding Western Australia’s highly regulated retail trading hours one of the chief economic benefits cited by advocates of deregulation has been that longer trading hours create jobs and reduce unemployment. Critics of the theory argue that the retail sector has already exhausted the potential labour pool and that relaxed regulation will simply divert workers from small businesses to larger and more aggressively competitive outlets. In November 2010 the WA government under the direction of Premier Collin Barnett introduced new legislation to lengthen the allowed opening hours for general retail stores by an additional 3 hours on weekdays, allowing shops to stay open until 9pm. One year later critics were quick to point out that the unemployment rate in WA was up by 22 basis points, while others responded that we fared well relative to the overall decline in employment across Australia during that same period, which was 7 basis points higher.

To make any conclusions about how the new trading hours legislation shaped the WA labour market some amount of further examination is required. In 2008 the Australian Bureau of Statistics released a special report on the condition of the retail labour market in WA, providing a basis for comparison in the 12-24 months before and after the new rules were enacted. The ABS also provides a detailed quarterly report on labour markets across Australia compared to which the performance of the WA labour market and the individual industries it comprises can be benchmarked. At time of writing the latest report covers the September-November quarter of 2011, providing a full year of employment data following the 2010 reform.

Taking the data at face value, as at November 2011 there were 3700 fewer workers employed in the retail sector than there were 12 months earlier. Relative to other industries in WA, retail lost half a percentage point of its formerly 10.5% market share, while nationally the retail labour market remained steady at 10.5%. This was not a smooth transition in WA: there was actually a sharp uptick of retail employment share in the six months following the change, obscured in the results by an equally sharp fall from the quarter preceding. This kind of cyclic behaviour makes trends difficult to identify, but in this case even smoothing out the cycles with seasonal adjustments yields, at best, no relative growth in WA’s retail sector since trading hours were extended.

Retail share of labour force

The post-reform declines in both number and relative size of WA’s retail labour market may appear to defy the clear reality that many shops are indeed open later, with supermarket chains such as Coles and Woolworths requiring many workers at each outlet to operate.  Identifying the source of this labour requires looking a little deeper, at the composition of employment within the retail sector.  ABS data reveals that the extended trading legislation came in the middle of a strong upwards trend in the proportion of retail workers employed full-time. While nationally the retail employment ratio was hovering around 51%, the proportion in WA was racing up from a low of 46% to a peak of over 54% working full-time.  Though the total number of retail workers had fallen after the change, there were on average nearly 2,000 more full-time workers during this period. The ABS figures show that full-time employees in retail work twice as many hours as part-time employees, indicating that the nominal loss in employment numbers experienced by the retail sector  may be substantially or entirely ameliorated by a large boost to the number of hours worked by the average employee.

Percentage of retail workers employed full-time

Instead of attracting new employees from the outside, when shops were allowed to open later they found the additional labour within their own ranks. Though the effect on unemployment in the state remains unclear, the increased availability of full-time work in the retail sector brings the benefits of greater and more regular income to the wide segment of society employed within in it. Retail in WA employs more workers than any other industry and plays an important role in setting prices for consumer purchases. In the year following the introduction of extended trading hours we have already seen extensive restructuring in the balance between full time and part time employment. The full impact may well be yet to come.