Budget honesty should mean incorporating distributional analysis


There’s an old saying: never cross a river because it is four feet deep on average. Similarly, a budget described only in averages or totals can hide staggering inequality.

Few issues in economic policy have generated as much discussion in recent years as inequality.

The IMF, for example, has repeatedly warned that rising inequality may detract from growth. US Chair of the Federal Reserve Janet Yellen has said inequality could damage equality of opportunity. In 2016 (then) US Treasury Secretary Jack Lew worried that rising inequality may undermine liberal democracy and free market capitalism. The OECD has also raised concerns.

Public interest in inequality is such that even an academic economist’s 700-page book on the topic became an international best seller.

And yet, you would never know it from the Commonwealth budget papers, the government’s principal statement on the economy and fiscal policy. In the 2017-18 Budget, the word “inequality” did not appear once in the 354-page Budget Paper 1, the paper for overarching economic and fiscal issues.

Inequality has only appeared twice in the past 10 years of Budget Paper 1 (3493 pages). Reference to Australian “poverty” appeared only once. It’s a similar story for related terms like “equality” or “equity” (excluding the accounting term).

Budgets could help inform public discussion about inequality by publishing “distributional analysis”, which is quantitative reporting about how people across the income spectrum are differentially affected, in this case by policy decisions in the budget.

Distributional analysis in the UK

One nation that publishes budget distributional analysis is the UK. Its 2017 budget, for example, includes a 21-page distributional analysis document, which reports on existing trends in society’s income distribution, and then projects the impact of the budget on households by each income decile. (Income decile refers to ranking households into ten groups, so that the 10% with the lowest incomes are in the first decile, and so on.)

The projections aim to capture the aggregate impact of all budget measures that materially affect households’ financial position, including changes to taxation, welfare and in-kind public services such as free health care. The net benefit or cost to them is shown both in pounds and as a percentage of household income for each decile. Also published are distributional projections that, in addition to the current budget, include all policy passed by the current parliament.

Click to enlarge.

There is a strong case for Australia to follow the UK’s example.

A thorough description of the budget

Aside from any moral considerations, distributional analysis is desirable simply because it better describes what is happening with the budget and economy. There’s an old saying: never cross a river because it is four feet deep on average. Similarly, thinking about a budget or economy only in averages or totals can be deeply misleading.

To borrow an idea from Anthony Atkinson, a leading inequality scholar until his death in January, a society in which some can afford space travel while others can’t afford food is profoundly different from a society in which everyone can afford to buy food albeit with no space travel, even if they have the same national income per capita.

Thus consideration “of the distributional dimension is necessary,” writes Atkinson, “if we are to relate the big numbers of economic policy – such as GDP – to the real-life experience of individual citizens.”

Similarly, the fourth recommendation of the 2008 Stiglitz-Sen-Fitoussi commission on measuring social and economic progress, headed up by two economics Nobel Prize winners, called for governments to place greater emphasis on distributional statistics.

Community values and preferences

Distributional analysis is desirable not just as a neutral, technical descriptor (although it is that), but also because it is relevant to prominent community preferences and values. There is considerable interest in who gets what from the budget, partly because most Australians would prefer a more equal society. This is consistent with a substantial literature of economic experiments, showing popular aversion to inequality.

Given such beliefs, mainstream public economists have long recognised distributional equity as one of the objectives a society wants from its economic policy. Two Nobel Prize winning economists, Samuelson and Arrow, advanced pioneering models in which social welfare is dependent on distribution. No fewer than four other Nobel Prize winning economists – Kuznets, Sen, Stiglitz and Krugman – have worked on inequality.

In a seminal 1975 book, economist Arthur Okun argued that the great task of economic policy was to balance equality and efficiency. This has become a public finance orthodoxy. Both the Henry Tax Review of 2010 and the Tax White Paper Discussion Paper of 2015 listed distributional equity, alongside efficiency, as goals. In 2009, Ken Henry said that to his knowledge this was consistent with every tax review conducted by a developed country in modern times.

It is remarkable, then, that in the government’s principal statement of economic and fiscal policy, distribution does not rate a serious mention.

Australia has capability

Treasury already has the modelling capability to undertake distributional analysis, and past governments have elected to publish its distributional work on major reforms, such as the GST and carbon price packages.

The Australian Government could adopt the UK’s decile analysis, or use quintiles (five income groups) as with previous work by Treasury. If quintiles are chosen, they should be augmented with the bottom decile. This reflects interest in the welfare of the poorest, following a variety of social, philosophical and religious traditions.

There should also be a top 1% category, given community concerns about plutocracy.

Incorporating economic projections, the modelling should show the expected real disposable income for households by income group, and show the change that is due to the budget (both in dollars and as a percentage of household income).

Distributional estimates for the previous year should also be released, with a gini coefficient, as this is useful to summarise overall inequality into a single index number.

Heading backwards?

Prior to the 2014-15 Budget, each budget’s overview document included “cameo analysis”, which is a simple form of distributional analysis. Cameo analysis projects change in the real disposable income of hypothetical people who vary on selected attributes, such as earnings, relationship status, parental status and so on.

Click to enlarge.

Freedom of information documents show that Treasury did undertake cameo analysis in 2014-15, and a negative effect on low income Australians was projected. Cameo tables have not appeared since.

Possible concerns

Why might policymakers not support publishing distributional analysis?

There may be a view that distributional analysis would shine a spotlight on the losers of a budget, making it harder to deliver efficiency-focused reforms.

While understandable, such concern is misplaced. First, Australia is after all a democracy and most of us like it that way. Questions about a community’s ideal income distribution are inherently value-laden, so they have no good answer independent of the values of the community itself.

Second, transparency improves governance on average. Like democracy, transparency could at times make certain reforms more difficult, but giving governments free rein can lead to much worse. Ultimately, if economic reforms are to be sustainable in a democracy, communities should be informed about the consequences and brought on board.

Third, budget distributional analysis might even aid the case for reforms, as it frames distributional concerns over a whole budget, rather than policy-by-policy. Opposition to reform typically exploits fear about individual measures – such as a GST or carbon price – ignoring the wider context.

For example, a carbon price is regressive, but due to substantial compensation the Rudd and Gillard carbon price packages turned out progressive.  And even though the GST is regressive, Howard’s GST package included progressive elements such as abolishing certain wholesale sales taxes and increasing government payments. Distributional analysis is what shows this.

Broader analysis, greater transparency and stronger democracy

Publishing distributional analysis would be good governance, because it provides the community with greater transparency on fiscal and economic policy. It improves the objective description of the budget, and relates to values that are fundamental to many in the community.

For these reasons, the Charter of Budget Honesty Act should be amended to mandate the inclusion of distributional analysis in the budget.

David Sligar has worked in the Commonwealth Treasury and Finance departments, primarily in budget and social policy areas. He has also worked as a parliamentary economic adviser, including for a former Australian Treasurer and a state Leader of the Opposition.