You can contact me directly here

This page is best viewed with a computer and a monitor

Support the Any Browser campaign and keep your freedom of choice on the Net.

WWW browser


Back to the home page.

These have been reformatted for the internet, and in some cases the printed form was edited further.

Advertisement scheduled to appear in the Australian and the Melbourne Age of 24.8.99

[Further information available in a letter among my Constitutional material]

Conduct of the Constitutional Centenary Foundation

Notice to members of the Victorian Chapter

Have you received the mail-out of 16 April 1999?

The CCF is meant to promote informed discussion on constitutional matters in the run up to 2001. This is especially important with the republic referendum coming up in November.

In 1997, at an extraordinary meeting, the Victorian Chapter Committee attempted to destroy completely and in secret the Summer 1997 newsletter which included articles on the Republic, and articles by Justice Howard Nathan and the former Northern Territory administrator, Austin Asche, amongst others, apparently in order to to suppress material showing that some republicans are not progressive.

That attempt failed. In March 1999, after legal proceedings had been commenced, the CCF agreed to distribute the newsletter to its members to let them judge the material for themselves in context. This mail-out appears to be dated 16 April 1999, but contains a notice referring to events alleged to have occurred at a meeting on 18 April 1999. Are you a member? Did you receive this newsletter? If not, please contact:-

563 Little Lonsdale Street
Melbourne, 3000
Tel: 9670 8951
Fax: 9670 2954

Return to the contents list.

Letter printed in the Australian Financial Review of 11.8.97

Tax rebate to create jobs

In his article of 6.8.97 Alan Mitchell summarises some theoretical options that were used to brief Ministers before the Cabinet meeting on unemployment. One in particular showed some possibility of results, namely lowering the minimum wage while compensating the less well off through social security and/or the tax system.

These options rely on reaching a new equilibrium. The practical difficulties are blindingly obvious - there's many a slip between cup and lip, so what if other factors interfered and we got the worst combination of the intended and current situations? We might have all the unpleasant effects of minimum wages collapsing, without ever quite reaching the jam promised us for tomorrow. We might not be able to get there from here.

Yet there is another approach. Consider the effect of rebating employers' own tax bills, say by $10,000 per full time employee per annum (and pro rata for part-timers). The precise figure should correspond to social security costs, including the on-costs of administration. The general tax rates would need to be higher to offset the rebates on the existing workforce, of course - but short of 100% unemployment the frighteningly high rates are merely dummy intermediate numbers for purposes of calculation, and no employer would ever pay that much. But this does imply broad-based production taxes, such as a GST or a tax on the use of commercial plant and premises.

In the short term such rebates would encourage employment at the same time as protecting a minimum wage corresponding to social security, with no additional cost over and above the existing hidden cost of social security. As tax rates would match present requirements the system would be revenue-neutral in the short term, and while the tax base would narrow to the extent that social security was privatised, it would only do so in exact step with reductions in the social security burden. Over time this approach would be budget-neutral and would not affect governmental planning, resources or budgets in any other respect. Yet apprenticeship schemes would become realistic, retrenchments would be less likely, and to the extent that the unemployed were taken into the workplace they would be exposed to all the opportunities they need. And, over time, structural change would still create new opportunities - only now without people having to risk being sidelined into dead-end, low wage careers.

With this approach you can get there from here.

Return to the contents list.

Letter printed in the Australian Financial Review of 7.11.97

Tax rebate on employees a better option [than negative income tax]

Professor Dawkins' article on unemployment and tax reform states that "the best way to attack these problems is with a negative income tax". Another approach uses broad based taxes on producers. If employers received rebates for all their employees, e.g. classifying them as a GST business input, there would be an incentive to hire at corresponding wage levels.

The only difference is where the tax falls. Yet this has three major consequences:-

Perhaps a drug cocktail attack on unemployment could combine these and other approaches.

Return to the contents list.

Unpublished letter to the Australian Financial Review, written 6.7.98

Time scale problems for economic reform ever reaching actual improvement

[these follow from mathematical considerations involved in, e.g., stability criteria for back propagation in neural networks or the convergence of perturbation methods in solving rigidly coupled systems of simultaneous differential equations]

You recently printed material on economic issues from Mark Latham (2.7.98), Ian MacFarlane (3.7.98) and Peter Costello (3.7.98). It may be misleading in isolation.

Mark Latham asserted that "In an era of permanent change, 'prepare for it' is the message that this Government should give"; he takes it for granted that permanent change should be accepted. But look at the implications.

Ian MacFarlane described how "The benefits of liberalisation take time... and in the intervening period some... will experience economic hardship... Liberalisation can cause a significant, though temporary, dislocation..." This is identifying short-term with temporary, forgetting the possibility of permanent change. It is only a valid engineering approximation that short-term shocks are temporary when they occur on a timescale about three times slower than assimilating them. If the road is always up for improvements, there is never any actual improvement.

Even temporary change may have permanent losers. Will new entry level opportunities go to teenagers or thirty year olds who never broke into work? Will the middle-aged retrenched five years ago reinvent themselves, or will those still in work get the first call? There could be permanent losers if we merely kept pace with change.

Peter Costello argued that "Australia just cannot turn its back on reform now - the financial markets... or... the IMF... would force it upon us". But with permanent change this dilemma of reforming or being reformed is itself dangerous; at best we can buy time and still be riding the same (Asian?) tiger. The reward for painful reform may be dependence on more of the same, forever or until we fail. Surely that is an underlying problem? Should we not aim beyond, either ending the reform process after a finite number of cycles, or if it must be permanent make it sustainable?

Return to the contents list.

Unpublished letter to the Australian Financial Review, written 2.5.99

Differences between Negative Payroll Tax and Wage Subsidies

I was privileged to be able to ask a question at the recent Business Council of Australia "New Directions" conference in Melbourne, about whether there had been any empirical studies directly on business tax credits linked to staffing levels rather than income tax credits. (These would generate a Negative Payroll Tax matching Social Security.)

Chris Richardson's reply was that he didn't think business tax credits would work, which I take it means no direct studies have actually been done and we must extrapolate from related work.

However most of that is in connection with Negative Income Tax or wage subsidies. A Negative Income Tax works differently by affecting demand, and takes time to flow through to the supply side. (So an NPT acts more rapidly and has no material initial costs.)

Wage subsidies generally fail because:-

Business tax credits are sustained and universal, which bypasses two problem areas. This would greatly increase the deadweight problem, except that they are differently financed. Rather than there being an outflow of government funds, there is a revenue shortfall which has to be made up by raising the general level of taxation. This does not increase the tax burden because employers are receiving the deadweight effect. At the end of the day the system is budget neutral to the government because it claws back the initial deadweight losses through the higher rates, and any further loss due to increased employment is offset by the lower burden on Social Security.

As someone pointed out at the conference, attaching business tax credits to a GST would mean a windfall gain to labour-intensive industries at the expense of capital-intensive ones. There are at least three solutions to this: direct subsidies to key industries; grandfathering in the tax credits by targetting them to people born after a cut-off date; and attaching them to a completely different tax. As this approach is against the spirit of a Payroll Tax anyway, I would suggest encouraging the States to replace Payroll Taxes with taxes on the use of commercial plant and premises which could carry the business tax credits. Such taxes could be fine-tuned to vary by industry sector.

Return to the contents list.

Letter printed in the Australian Financial Review of 3.10.00

John Hewson's non sequitur of 29.9.00

You describe John Hewson's article "A republic is the next event" of 29.9.00 as "Our overt nationalism at the Games further demonstrates why we should be a republic". This is pretty much the argument mostly put for a republic before the referendum, and he supports it in much the same way, with a brightly coloured montage of unrelated and emotive items. Only the headline, the opening paragraph, the byline and the punchline have anything to do with a republic at all, and to put this as an argument for a republic is a mere association of images. Indeed, if we can celebrate national pride without first becoming a republic, if we can conduct the Olympics with our present system without an overwhelming sense of shame, it serves to undercut any republican argument as necessary to secure self-respect.

It did not follow that Australia had to become a republic last November, in the face of actual and substantive issues that really were wrapped up in a republic, and nor does it follow now for all John Hewson's undoubted expertise as an economist or his experience of the political trees that obscure the constitutional wood. Non sequitur - it does not follow.

Return to the contents list.

Letter printed in the Australian Financial Review of 5.10.00

Alan Mitchell's article of 4.10.00

In Alan Mitchell's article of 4.10.00, "Freer trade helps the poor", it appears that throughout he is citing analysis of what happens to poor countries as a whole and not what happens to the poor within the countries. Even after comparing "...the potential gains from freer trade with the potential cost of imposing higher labour standards on developing countries", he goes on to bring out the aggregate benefits to those countries as a whole.

Is this in fact what he is answering? Because if so it goes very little way towards addressing what increased trade liberalisation actually does for the poor. It is true that wealth eventually trickles down, but this is a slow process, and without that the effect may be like the old saying about foreign aid: taking from poor people in rich countries to give to rich people in poor countries. It may be said that it only happens when there is something kleptocratic about the developing countries, but this is like saying "guns don't kill people, people kill people" - it begs the question of the reasonably foreseeable consequences of giving the means to those who are likely to use them that way.

Indeed, there are mechanisms that can cause some people to regress during the early stages of adjustment in any country, such as increased rents or increased use of land for the cash economy. Although it does not always happen, when it does it gives the poor a sort of J curve of progress, reculer pour mieux sauter. The trouble is that with several stages of change piling up one after another, the separate J curves combine to produce a downward trend, at any rate over any period that matters. So, regardless of the fact that the country is developing, we still cannot be sure that we really are helping the poor that way. We certainly cannot rule out these possibilities in advance from empirical experience - even our own experience shows people who have remained caught for years in the eddies produced by our own advancement, despite the fact that anyone not caught definitely benefits.

Return to the contents list.

Letter printed in Quadrant of March, 2002

Identity cards

You say various things about identity cards in your November editorial. Not all of them are borne out by experience, not even by the examples you give.

For instance, there is indeed a valid argument from expense against them. What counts is not the fact that a universal card would be cheaper than a multiplicity; the costs we now face are sunk costs, and a new universal card would impose new and additional costs.

As for risks of being on a central registry being "fantasy and paranoia", even the limited system we now have has already presented me personally with a real and continuing burden. As I am one of the unexpectedly frequent victims of faulty data matching, I have a continuing struggle to prevent being linked to other people of the same name and date of birth. To say that errors "can be dealt with easily enough", that is not so even now. That is, they can be dealt with but not easily; after two years I found that I had not received driving licence correspondence as it had been sent to Koo Wee Rup. As, when and if there is a universal system, far from these independent checks being as convenient the errors that creep in will be locked in; any "check" will simply be referred back to the central registry which will accurately if spuriously return the false information it holds. If on immigrating I found it hard enough to get a Tax File Number on the grounds that I already had one and lived in Gippsland, what would happen to people in my predicament in the future? The worst that has happened to me so far was being given a difficult deadline to challenge a Social Security penalty some years ago, with a contact address for an office that was in the process of closing, when one of my alter egos had left the country without my (naturally) telling the government. At least it is unlikely that I will face the fate of the person who was recently deported because of mistaken identity (page 6 of the Australian of 4.12.01), but if that ever happened to anyone just how easily could it be challenged? It was sheer luck that brought that case to light.

And that is just an example of what can happen to individuals when mistakes are made. It cannot be measured by any aggregate, as those tests simply slide over the problem. How much worse could they get if, indeed, there ever were bad faith at any level, whether institutional or from the misbehaviour of some individual? Both possibilities exist, with the latter being realistic if we can judge from recent court cases against a former ATO officer.

No, in this matter, as in matters of constitutional change, I prefer Henry Ford's design philosophy: when asked why his cars did not have a certain common feature, he replied that if he didn't put one in it couldn't break and it couldn't fall off. And, he might have added, the savings in money, space, and weight could be devoted to the other parts. In the days when automotive matters were untried, unproven, and often erratic, this was a sound principle whatever the advantages of the feature. Since - contrary to your position, and as I have found from my own direct experience - we do have serious defects in our existing identification approaches, I submit that this is the line we should take here.

Return to the contents list.

Letter printed in New Scientist of 20.4.02

"Land and Freedom" article

(This was somewhat edited before being printed. This version shows the original draft.)

The article "Land and Freedom" on page 17 of New Scientist of 23.3.02, and the fuller version at your website [or here], present differing views about whether these developments will actually help people they displace. The developers claim that they will find other jobs, and will be better off as there will be more GDP for them to share in.

There are several problems with this, pretty much the same as happened with the Enclosures in England or the Highland Clearances in Scotland but with one further aggravating feature:-

The only way forward is likely to be impractical: it requires compensation to the evictees, not merely to make sure they get a net percentage of the gain considered as a capital lump sum, but in the form of an indefinitely sustained subsidy flow that is enough for them to compete with other unskilled Indian peasants indefinitely even if not enough for them to survive on by itself. Otherwise they will be in the same unemployable boat as unskilled Australians who cannot compete with third world semi-subsistence labour without a similar subsidy or the equivalent, only without our Social Security safety net to keep them alive. (The Australian situation would be bad anyway, even without subsidised foreign competition - there is an externality involved in resourcing Social Security that unnecessarily raises levels of unemployment here.)

Return to the contents list.

Unpublished letter to the Spectator, written 15.9.02

Andrew Kenny's article of 7.9.02

Andrew Kenny's article of 7.9.02 comes to the wrong conclusion on some important economic issues, not so much from getting his material wrong - it's correct, as far as it goes - but from leaving out some important things that he couldn't be expected to know. But he should know to some extent - the very approaches he recommends as having helped the west, did indeed have horrendous side-effects in each of England, Ireland and Scotland in various ways in the 18th and 19th centuries. This, despite the fact that the aggregate gains were real and there are only beneficiaries left now to tell the tale.

Take his advice that developing country "small" farmers should increase their cotton crop with genetically modified seeds. There's no world shortage of cotton, so although these particular farmers would end up better off (at first), very little more cotton would be sold; other cotton growers would end up worse off, and a lot of the gain would go to buyers in developed countries from lower prices. After a while, from competition, all the growers would be forced into using the seed, so everybody would be back almost where they started. But they would all be paying for the GM technology, with no compensating local gain for the investment they would be servicing.

This payment overseas is an analogue of the absentee landlord scenario that Nassau Senior analysed for Ireland, which is disproportionately harmful in underdeveloped countries. But it gets worse. To the extent that land did get diverted from growing food into producing cotton as an export cash crop, other people in those countries would suffer. That includes the genuinely small farmers, subsistence farmers who end up dispossessed, not just the employees of the comparatively wealthy cotton growers and the urban poor, because it's only the growers who decide what's profitable. And no, the marginalised don't end up paying for imported food with their new found wealth; they don't have any.

So far this is just what's typical, for instance for coffee in most countries that grow that, but it gets worse. Cotton is a thirsty crop; whereas owner farmers can decide for themselves what pays best, using water resources has a cost that spills over onto the rest of the farmers. And these countries don't have proper water use systems, and if they ever did get any then running those would cost as well.

This whole area is crawling with market imperfections that prevent the gains reaching the poor, not least the way the dispossessed can't move to the areas where the gains are occurring - those are in the countries where they get their shirts cheaper.

No, as Peter Jones may tell you, the kind of reforms that are likely to help are those that Solon and the Athenian tyrants tried with some success: ban exports of staples, confer security of tenure on land use to encourage improvements, and only permit exports of high value added products and agricultural products of a sort that don't compete with food production for resources. In Athens they exported olive oil; what is today's equivalent? Not cotton.

Return to the contents list.

Unpublished letter to the Melbourne Age, written 1.1.03

Geoffrey Heard's letter of 30.12.02

Mr. Heard claims that English experience of common land shows that English law recognised and worked with "community rights of property".

In fact the latest newsletter from "Prosper Australia" has an article proving it worked precisely the opposite of that, so John Dawson was right after all. On page 15 the newsletter states that "contrary to a widespread belief, all common land is private property", citing sources.

The thing is, common land did not embody any idea of a "community" that would have been an entity owning and controlling land. Rather, for each village there was a restricted group of commoners who each individually had certain restricted rights on common land, say to graze animals or collect fuel wood. The public at large was barred, even commoners from other villages and underprivileged people from the same village.

What this shows is that even when there was a community, English law had to rearrange the problem so it could be handled one on one by individuals. Common land doesn't mean any legal precedent for handling land rights communally after all.

Return to the contents list.

Letter printed in the Spectator of 1.2.03

Bunyan in Oz

Peter Hitchens asserts [in an article of 18.1.03] that Philip Pullman is writing a sort of inversion of C.S.Lewis' Narnia books, tracking them but inverting values and incidents that work their way out in the books and so producing something atheist where Lewis was Christian.

This is hardly the first time something like this has happened. In the nineteenth century a clear pastiche of Bunyan's Pilgrim's Progress appeared, with a journey that ended not in heaven but in finding and taking the way back from the very destination that Bunyan warned was there. It even inverted identifiable incidents, like Christian's ingenious way of avoiding being destroyed by the glory of heaven before he was ready for it. I am amazed that this obvious antiparallel was not seen for what it was and banned by every library in the USA.

I refer, of course, to the Wizard of Oz.

Return to the contents list.

Unpublished letter to the Australian Financial Review, written 8.4.03

The Liberal Party's practical help for the disadvantaged

In his article of 7.4.03 Geoffrey Barker calls on "liberal conservatives" to remember the less well off rather than pushing them down while the rest advance. He also reminds the "ideological" liberals that they still have a selfish interest in this - a point Disraeli made before.

The thing is, right wing parties have a vested selfish interest in this sort of help; promoting people out of poverty increases their constituency. And, of course, left wing parties have no interest whatsoever in that - all the poor can reasonably hope for from them is a drip feed sufficient to maintain a captive constituency in its captivity.

But the Liberal Party of this country has not forgotten, still has those who care without being fuzzy. At the most recent Victorian Liberal Party State Council the Beaumaris branch put up a motion on this theme, and at the one before my own branch (St. Kilda Road branch) got a motion passed to have Professor Kim Swales' work in the area studied - work which suggests a helpful market reform. The fact that in six months nothing has been done about it says more about the party's priorities, resources and capacities than about whether Disraeli's message has been forgotten.

Unpublished letter to the Australian Financial Review, written 22.4.03

Peter Saunders' and Andrew McCallum's article in the Australian Financial Review of 17-21.4.03

In their article of 17-21.4.03 Peter Saunders and Andrew McCallum present two opposite poles of a debate on how to help unemployed people re-enter the work force.

Even if we confine ourselves to the two poles on offer, there is a considerable amount at cross purposes. Neither description allows enough for time scale issues and their transitional costs, which amount to an uncertain investment. Increasing minimum wages does indeed cause the harm Peter Saunders fears; only, not straight away - and in the short term there really would be the improvements Andrew McCallum wants, before their damage exceeds their gain. That makes a vicious spiral of dependency.

Yet, precisely because we are already in that spiral, doing the opposite would first produce the harm Andrew McCallum fears; only later would there be any improvement. Effectively, we have already drawn down and lived off capital that was embedded in a healthy full employment economy. It would cost us to reinvest that capital. And unfortunately, for complex reasons to do with how advanced our economy is, it is unlikely that any simple relaxing of the burdens would be enough. It is more likely that raising tax thresholds would eventually improve the lot of many but still leave others marginalised - yet more inequity. Even the gainers would have to wait for the working environment to change before they got anything.

Peter Saunders accurately states that raising personal tax thresholds is "the most radical solution on the table". That is a large part of the problem - other candidate solutions like Professor Kim Swales' GST tax credits just aren't on the table yet. I know Peter Saunders knows about that one, because I myself described it to him when he was recently in Melbourne. Yet these candidate solutions would bypass all that delay and costly investment in improvement. They are at least valuable as part of a transition.

We shouldn't polarise the debate but look for wider options and areas of synthesis.

Return to the contents list.

Unpublished letter to the Australian, written 9.5.03

Max Boot's article in the Australian of 8.5.03

In his article of 8.5.03 Max Boot makes out that "on the whole US imperialism has been the greatest force for good in the world during the past century."

This is manifest nonsense. Even if we grant that imperialism can ever do net good, that it makes sense to net these things off in that way, there were already European empires trying to do precisely that right up until the middle of the last century. It was US efforts that stopped them, so even if there was good done it was at the expense of good that the USA prevented; so the USA cannot possibly have done any good on the whole. And if there was not good done, if it was wrong then of (say) Britain and France and Israel to try to stop dictators from coming into being in the Middle East because there was also something in it for them at other people's expense - why, it is at least as wrong now of the USA to try to remove those dictators at that price without either the Europeans' hard won advantage of experience or their moral high ground of never encouraging the dictators.

As far as the good of imperialism goes either we owe the USA nothing since it took more with one hand than it gave with the other, or we owe it nothing since it wasn't doing good things at all.

Return to the contents list.

Letter printed in the Sunday Age of 25.5.03

US ponderables

(This was somewhat edited. As some of the cut material was significant, this version shows the original draft with the cuts left in. They are roughly indicated by being coloured red. This site includes the version as printed.)

In his letter of 18.5.03 Ilbert Phillips makes a number of [several] pro-US statements, backing them up with a whole series of misunderstandings. Let's go through them:-

Return to the contents list.

Unpublished letter to the Spectator, written 11.6.03

Johan Norberg's article and Banned Wagon of 7.6.03

In the latest Spectator, Johan Norberg has told us roughly the right thing for the wrong reason - and similar reasoning in Banned Wagon has pointed in quite the wrong direction.

Yes, all the things Johan Norberg states about benefits from globalisation to workers in third world countries are correct. Only, that has very little to do with whether that is helping. The catch is, the workers are not the ones at the bottom of the heap in those countries as the poor there aren't workers yet - and without all sorts of proper markets, matching property rights and good governance, the marginalised don't end up being promoted into being beneficiaries any more than they did with us during the Highland Clearances or the Enclosure of the Commons. And countries without proper markets, property rights and good governance are the kind there are.

What Johan Norberg did not look at was what new Nike factories cost the people who don't work there. As it happens, that's practically nothing - so his conclusions are correct overall despite his reasoning being incomplete, and Nike is indeed a benefactor there. But it's another story with exported cash crops like rice, coffee and cotton, which he touched on and which were the thrust of Banned Wagon.

What happens with those is just what that notorious dry Nassau Senior analysed in the 1830s, when he showed that absentee landlords did not hurt manufacturing England but did hurt agricultural Ireland. A multiplier means that people there needed to export even more and ate into their own subsistence requirements; the cost of living rose disproportionately, even as non-cash subsistence dwindled. So paid workers and cash crop farmers, compradores and cottiers, do indeed get better off - but the peripheral workers start to starve. Yes, their cash incomes rise, but their subsistence resources vanish; it's like the story of Ruth and Boaz, only with Boaz not even knowing that Ruth the gleaner would perish as he ceased to leave gleanings while he grew rich with new ways.

This ought not to happen, but it does: because the marginalised don't have property rights in the resources they used to have, and so aren't compensated for their loss (poor institutions in those countries); because there are kleptocrats there, not all being like Boaz but rather like the rich man in the parable who took the poor man's ewe lamb (poor governance); and because of something from outside, from our new world order. That something is the way we don't really compensate these countries for the change.

What ought to happen, roughly what happened when Britain invested overseas a century ago, is that real funds flow out and make new roles appear. Nassau Senior's nasty scenario never happened then, because absentee ownership of foreign investments was matched by making those investments in the first place - as those funds went overseas they made the very new opportunities that were needed. The Argentine ended up with railways as well as exporting beef, and nobody starved from losing subsistence because they found new jobs arising.

But now is different. Now the US dollar is the world reserve currency, and as it is a fiat currency a large part of the funding corresponds to printing new dollars and isn't really making new investments at all; it's just mobilising local resources and acquiring existing investments. This has the effect of moving food from mouth to mouth and not making new opportunities at all - you can't simply say the poor ought to take the Nike jobs, as not enough appear. If you apply Johan Norberg's reasoning to these countries now, it all looks good as this or that peasant makes good - but it leaves the marginalised out of the accounting, as they get squeezed off the edges with nowhere new to go.

It is a sad final irony that charitable groups are even now agitating for these countries to be thrown open to wider agricultural export opportunities. As things are, without first arranging for those new jobs to arise in step, that can only condemn the poorest of the poor to even worse.

Return to the contents list.

Unpublished letter to the Spectator, written 1.9.03

Anthony Daniels' article of 30.8.03

Anthony Daniels compares Pauline Hanson with Jean-Marie Le Pen; a better comparison might be with the late Pierre Poujade.

Hansonites are a broad church, not a racist one. The National Republican Movement doesn't make all Australian republicans into racists. In fact Hanson's offence consisted in measures to prevent racists and the like taking over. Australians' real objections to greater migration are merely being demonised as racist; but they are only objections to more people than we can digest and provide infrastructure for, leaks that gain on the pumps. Any cultural preference for (say) the British is not prejudice but acknowledgment that some people have a cultural head start in being easy to digest. There is no objection to giving others the same head start as well, just to throwing too many at us too fast and too far off the deep end to cope with. That just drags others down too.

There is no great irony in Hanson deploring light sentences yet not getting one. The main objection is one of principle, to the conviction being "proper" at all, to laws that make such things a crime in the first place. This is how Pauline Hanson compares with Nelson Mandela, not in what they stood for but in the way they really were handled under due process of law - it's that and not Hanson's actions that reduce our confidence in the electoral process. It says more about the system that makes the law, and which the law supports, than about the people caught under it or brushed aside by it. (The long established Democratic Labor Party is being brushed aside right now.) It is worse than a crime, it is a mistake to sit on the safety valve of democracy. Australian democracy enforces a party system as an intermediary priesthood, one that filters out and distances the people, so Hanson needed one too to be in the game. Staying out of the kitchen is a solution to getting too hot, but not to stopping the cooks running things hot throughout the house.

The underlying problem is institutional. Over the decades well meant things like secret ballots and paid representatives were followed by compulsory voting, direct public funding of parties, and compulsions on representatives not to abstain. We even have some of the targeted exemptions to compulsory voting that Aristotle condemned as a fraud on the public. Grass roots support became unimportant with these gentleman's agreements between the parties, and the insider culture changed imperceptibly slowly away from the people. So now we only have isolated insiders, and party structures are barriers. I have experienced this myself: over a year ago I got our (right wing) Liberal Party to resolve to study Professor Kim Swales' approach to unemployment - not implement it, merely study it. The party has not done so, and will not provide me with contact lists, letterheads, or other authority to do it myself. They do not go in themselves but stop others from going in. Hansonites cannot simply work constructively within the system - it's closed to new entrants. Even when the politicians think they are listening, it's only within a groupthink round robin of politics, the media, and academia. I have more chance of contacting the Australian public through the pages of the Spectator than through anything here.

By the bye, this elite of ours is not famous for drinking Champagne but Chardonnay. If they were closer to the real world they might not. I once knew someone who refused to touch it because he had worked on a farm during a mouse plague and had fallen in the grain enough times to know whereof he spoke. He said Chardonnay tasted of mouse piss.

Return to the contents list.

Unpublished letter to the Australian Financial Review, written 14.9.03

Peter Saunders' article in the Australian Financial Review of 12.9.03

In his article of 12.9.03 Peter Saunders is solving a problem we haven't got, how to get the long term unemployed looking for additional jobs - ones that they wouldn't just be taking from someone else. He suggests that cutting payments to the long term unemployed will encourage them to find jobs rather than slumping into despair, but he supposes that this pressure will not be deadening itself and that it won't itself eat into even the job vacancies we have now.

We aren't facing a situation with employers crying out for workers only to find the long term unemployed unwilling to become employable. At the moment the regular supply of new vacancies really is met by the oversupply of unemployed, and employers do fill all jobs.

While the ratio of seven or eight unemployed for every job doesn't mean there aren't jobs along eventually, the preference of employers for fresh and skilled workers means opportunities get mopped up before the long term unemployed get a look in. Peter Saunders' hope is to change this. But at best that could only make the long term unemployed more eager to take work than those not out of work so long - it would be a substitution effect, poaching jobs that would have gone to fresher people. It wouldn't even work if the same pressures faced those fresh people, motivating them too.

And what if new "jobs" people became willing to take under pressure were antisocial ones, from petty crime on down? Social Security was invented to head off problems like that.

To see the absurdity, consider this improvement: what would happen if we forbade employers to hire people until they had been out of work for a long time? It would clearly be unjust, and it would mean less multiplier increase in GDP to draw in workers since the ones on offer would be staler. With less GDP there would be more unemployment all up. But it would have one thing going for it, at least employers would try to offer retraining they themselves valued instead of the present sort which they don't.

No, to be truly constructive we should try the "five economists' plan" or Professor Kim Swales' similar approach using GST that would actually increase employment and not just move it around. Peter Saunders' idea can only move problems from one budget area to another, fooling people that there was an improvement without ever getting anybody a job that wouldn't have gone to someone anyway.

Return to the contents list.

Unpublished letter to the Australian Financial Review, written 6.11.03

John Quiggin's article in the Australian Financial Review of 6.11.03

In his article of 6.11.03 John Quiggin states that "by giving up on the idea of the governor-general as a neutral figurehead, the government has effectively conceded defeat on the main arguments against a republic and direct election."

Even supposing that the idea has indeed been given up on, rather than having a brief shadow cast on it, that has nothing to do with the main arguments anyway - although it does have a lot to do with republicans' own objections to a direct election model. However if Professor Quiggin or anyone else wants to hear the main arguments against a republic in general, rather than the republicans' own arguments against one particular model of republic, they shouldn't look for them from republicans.

I myself know of several general arguments against, ones that don't merely apply to this or that model. Here is one example. We know by simple observation that there are far fewer successful long term republics than monarchies - just the USA and Switzerland, and not Canada and Sweden as a republican once suggested to me. We don't know what the problems are, but we do know that there are problems and that it is for republicans to find a way past rather than require the status quo to justify itself. We don't know what led to those two exceptional successes but we do know that it may well have involved each passing through the fire. Taking all in all, we don't know how to repeat the trick let alone how to avoid that intolerable cost. Yes, monarchies also paid similar costs - but ours are sunk costs that happened far away and long ago.

Or to put it another way, if it ain't broke don't fix it.

Return to the contents list.

Letter printed in the Australian Financial Review of 6.1.04

Hegemony a first step to imperialism

In his article of 31.12.03 Professor Quiggin has no doubt inadvertently gone off on a false trail.

To give just one example, when he writes that a "characteristic feature of empire is that attempt to extend a single model of government and a single concept of citizenship over many different peoples", it just isn't true about empires in general. It's what the French did, but that was just what the French wanted to do with their empire to live up to their own justification to themselves, their "mission civilisatrice".

Most empires were multicultural and cosmopolitan. The Romans, the Turks, the British - all these didn't care about moulding the people they ruled, and they even provided institutional ways of coping with differences. In India the British kept so far hands off that as much as a third of India remained under indirect rule and following local practices right up until independence. In fact it was independent India and Pakistan that started trying to move in deeper. That is what made Hyderabad and Kashmir what they are today, not any heavy handed British monoculture.

So empires did actually use the methods of hegemony. They only tended to move into more direct rule when indirect rule came apart in their hands. There is no contradiction between hegemony and empire at all, and to show that the USA is being hegemonic isn't showing that it is not an empire, just that it might only be on the first steps towards empire. US anti-imperial intentions don't disprove it either, since even Britain didn't develop imperial ambitions until after it acquired an empire "in a fit of absence of mind".

Return to the contents list.

Unpublished letter to the Australian Financial Review, written 14.1.04

Andrew Leigh's article in the Australian Financial Review of 14.1.04

In his article of 14.1.04 Andrew Leigh concludes from Western Australian experience that a 1 per cent rise in the minimum wage is associated with a 0.13 per cent fall in employment, a small and acceptable cost.

There is good news and bad news about these results. The bad news is that they have not looked at the right measures. Since an initial fall in employment in Western Australia leads to a fall in tax revenue there and an increase in Social Security outgoings there funded by revenues raised in all states, there are external costs. Some of the burden is spilled over onto other states; a proper methodology needs to reveal adverse employment effects across all states. Just reporting the local adverse effects only tells us a lower bound on them, and there is some reason to suspect that the true figure is materially larger.

The good news is that what drives all this isn't really the minimum wage but the marginal cost of hiring. Standard methods of dealing with external costs can raise the minimum wage without flowing through to the marginal cost of hiring. Senator John Cherry of the Democrats has in effect just suggested one method, using tax credits on income tax. This would amount to a "Pigovian subsidy" to eliminate the external costs. Unfortunately it has costs of its own, mostly from time lags and from being somewhat targeted, but other methods like Professor Kim Swales' tax credits on GST even avoid those costs.

Return to the contents list.

Unpublished letter to the Australian Financial Review, written 21.1.04

US marijuana taxes not for revenue after all

Contrary to what Graham McCorry supposes in his letter of 21.1.04, the USA did not use 1930s taxes specifically on marijuana for revenue. Rather, using federal taxing powers was a legal trick to circumvent the US constitution which did not let the federal government ban marijuana. The tax law was passed and tax stamps were printed up to comply with legal requirements - but none were ever issued administratively, so making all actual marijuana sales illegal.

Of course this sort of lifting with the back misleads casual observers into learning the wrong lessons about the effectiveness of different measures. But worse still it reduced the moral force that law is supposed to have, as well as violating the spirit of the US constitution. This was a triumph of the tacticians over the strategists, not a victory for either integrity or utility.

Return to the contents list.

Article circulated privately and written on 30.3.01

Nine economic options for Australia

N.B., Some of these measures require others as preliminaries or as long term adjustments.

  1. Short term. Give farmers grants to convert diesel engines to use biodiesel, and loans to establish biodiesel processing cooperatives in the regions. Undertake to buy Australian grown biodiesel for government purposes.

    Advantages: helps the rural sector and the balance of payments in a sustained way.

    Disadvantages: direct cost and opportunity costs of distorted agricultural sector (but the distortion will clear as fossil fuels become uneconomic).

  2. Long term. Promote renewable fuels with tax breaks, including renewable alcohol/vegetable oil two stroke fuels which actually improve net CO2 emission levels by locking up biological carbon in soot.

    Advantages: helps the rural sector and the balance of payments in a sustained way, and helps with environmental obligations.

    Disadvantages: opportunity costs of distorted agricultural sector and foregone tax revenue (which also clear over time).

  3. Short term. GST and BAS complications can be headed off by "composition". Small businesses (defined by low turnover and by not being diversified) should be offered a substituted form of payment, based on a proportion of turnover. This proportion should depend on the category of the business, worked out by the bureaucracy from existing ABS figures to yield the same as the equivalent GST. (Clearly this is only possible when there is only one whole or main category of business activity.) Compliance should be monitored by the government, using modern statistical methods on the quarterly returns and initial statutory declarations about business category. Businesses dealing with these compounded payers would work with deemed embedded GST payments, to get their own business credit entitlements.

    Advantages: reduced compliance costs.

    Disadvantages: none, unless you want to increase bureaucratic monitoring for its own sake.

  4. Short term. Employment can be boosted by allowing GST and compounded GST payers a discount of some $10,000 per full time employee per year (pro rata for part timers, based on a 40 hour week). This boosts employment but causes a yield shortfall for the States who will be getting that revenue - but they don't have a short term problem anyway, and they can means test free services until employment improves, then cut them. Fiscal drag can be allowed to claw back some of the value of the GST offsets, eventually leaving them at some 5% of average wages.

    Advantages: improved employment and GDP.

    Disadvantages: the yield shortfall that needs to be made up, and a risk to the balance of trade from higher levels of demand for imports.

  5. Short term. Apply a temporary revenue tariff (ideally to exports, and best of all to exports of primary products) to bridge the temporary GST shortfall. Hedge the currency changes, but only with respect to forward purchases of fuel. Because of impact/incidence issues, this and the previous measure cancel out in most respects but still promote employment.

    Advantages: makes up the yield shortfall and also chokes off the risk to the balance of trade, minimising harm to servicing foreign debt if primary products are targetted.

    Disadvantages: might be misunderstood as protectionist, particularly if prolonged; may need careful explanation of how the measures do not cancel completely.

  6. Long term. Either raise the GST rates to compensate for the yield shortfall, or (better) the States should provide a specialised Land Tax to make up the difference, falling only on commercial premises and deriving tax rates from average profit levels of businesses in each tax area (so not rising materially with each business's own improvements). This also permits adjustments for the windfall gains of labour intensive sectors without harming the overall boost to employment.

    Advantages: makes up the yield shortfall without harming either the employment gain or the GDP gain (in percentage terms, this may be about half the employment gain).

    Disadvantages: releases demand for imports, which requires a boost in saving to compensate; merely raising general GST rates may lead to excessive marginal costs to businesses.

  7. Short term. Apply selective Income Tax cuts by age, with over 55s to pay a simplified flat rate of 20%. This boosts savings, and will encourage self-funded retirees to migrate. Compensate for the free ride to the superannuation industry by requiring it to transfer existing low-risk revenue yielding assets to an Australia Permanent Fund, according to actuarial projections. The returns from this compensate for the loss of revenue.

    Advantages: improves the balance of payments by allowing these assets to be drawn down, and by bringing in the funds of retirees.

    Disadvantages: the advantage is unsustainable, and such a fund is exposed to sovereign risk if a tax burden is reimposed on those who already compensated for their taxes by releasing these assets.

  8. Long term. Entrench the Australia Permanent Fund constitutionally, to prevent it being run down (compare the Alaska Permanent Fund and Alberta Heritage Fund); also require a prudent proportion of any deficit to go to the fund. Warn the electorate that the ALP cannot resist that sort of accessible resource.

    Advantages: eliminates the sovereign risk.

    Disadvantages: restricts the government's freedom of action, effectively hypothecating revenue.

  9. Long term. Move back pensionable ages by one year for every two calendar years that pass, lowering the cutoff age for the Income Tax break to match actuarially in compensation. It is important to prevent the ATO deeming retained profits in businesses to be notional income of owners, withheld to put the income into a year after the cutoff - such saving is useful. Require the superannuation industry to shift its administration and assets gradually to the Australia Permanent Fund.

    Advantages: builds up long term savings, effectively switching unfunded liabilities to funded liabilities.

    Disadvantages: slow, and further restricts the government's freedom of action (but acceptable as previous measures reduce people's dependence on the provision of services by governments).

I am indebted to the work of Professor Kim Swales of the University of Strathclyde and his colleagues for certain statistical details, and to Dr. Tim Hendtlass of Swinburne University for details of modern statistical monitoring with neural networks.

Return to the contents list.

Determination of the 135th Victorian State Council of the Liberal Party, 12.10.02 (resolution no. 23, submitted by the St. Kilda Road branch)


That a small working group be set up to evaluate Professor Kim Swales' approach to helping unemployment and to report back on it in an Australian context.

[Despite the very specific resolution passed by the State Council, as at 1.10.07 action is still bogged down in the Liberal Party bureaucracy, which so far has only mistakenly referred the matter to the Australian Public Service for a ruling on the desirability of one particular one of Professor Kim Swales' suggested policies instead of setting up a working group to look into his whole approach; it is hardly necessary to add that the Public Service has even misunderstood what is involved in that one policy.]

Accompanying statement

Several research groups have offered a number of approaches to helping unemployment. One of these is the "five economists' plan" for a variation of Negative Income Tax. While promising, it has serious continuing funding problems and also has up front costs to be carried until it starts paying back. Professor Kim Swales of the University of Strathclyde in Scotland has designed a variant that makes use of GST rather than Income Tax and which gives employers a tax break rather than assisting workers with a labour subsidy. It is constructed to avoid these particular problems, at least in the UK. We should look into what it does instead and find out any side effects, then see if it can be worked up into practical Australian policy within a wider package.

Executive Summary (given to St. Kilda Road branch)

The "five economist's plan" is a variant of Negative Income Tax, as is the American "Earned Income Tax Credit". This replaces a collection of support systems with a single basic one, working through Income Tax; for people on low incomes there is actually a payment, and for people with no other income they just get the equivalent of Social Security this way.

The thinking is that by reducing the wages that potential employees have to hold out for, and by eliminating poverty traps, it will become practical for employers to offer lower wages that are still realistic enough for everybody to price themselves into work.

The known catches are:-

One particular feature is that this has losers as well as winners, and it gets the losers early while winners don't turn up until much later. The hope is that in the end everyone ends up better off - maybe a few elections down the track.

It's possible to aim at the same general target and achieve even worse results; that's what the Greens' idea of a "Guaranteed Adequate Income" would do, since it needs far more funds for far longer (impossibly huge funds, in fact). EITC compromises on moving the poverty traps around, and also minimises funds outflow during start up (the unemployed don't get any funds - which also means they can't take just any job, since they need to finance [surviving during] their first work from it).

Professor Kim Swales' work eliminates all or most of the problems. This is by using GST rather than Income Tax. Under this, employers get what amounts to a Negative Payroll Tax, with their GST bills being reduced by a standard amount per full time worker they employ; this can be set as high as Social Security, though long run levels can be lower and you can start with the low levels if you do not need quite as rapid results.

Again, in the standard version the tax take is kept up by increasing the nominal rate of the carrying tax, GST. In a model using the British equivalent, VAT, Professor Swales found that an NPT set at 5% of average UK wages would lead to serious improvements in both employment and GDP, under a number of varying assumptions. You can find his material within at [now here]. I have also analysed this using Game Theory (see the link in my signature below).

The improvements are in these areas:-

There is still a problem from higher tax rates, but this is mostly illusory since all actual payers - employers - are getting their adjustments before they pass the whole burden on to consumers, and the adjustments can be set lower than with Negative Income Tax. However there is still a real problem, since Australian GST is a defective consumption tax. Without going into too many details, this can be cured in a number of ways. It can be patched by having the States make up the shortfall themselves with an increase in Land Tax on commercial properties (but not by increasing Payroll Tax, which harms employment by undoing all the good work of NPT). This works because they get all the GST revenue, they can apply the patch, the patch also falls on businesses that pass it on to consumers, and the burden of the patch does not vary with the amount of business done (which was the heart of the problem). And it's politically realistic since the States are currently under different management, and Federal promises to remit the States all the GST revenue don't apply to keeping the GST revenue up.

It also seems that there could be a problem "at the edges", where one tax regime meets another - which means, with foreign trade. So we have to be very careful looking into all the ramifications.

[The signature on the original email led here.]

Return to the contents list.

Article printed in News Weekly of 18.4.98

Today's unemployment and some of the cracks rational economics falls through

In an interview the other day Bishop Challen wondered how the world could reward company advisers for downsizing. The big puzzle is how the numbers work out so that we are all somehow better off for putting people out of work. Rational economics jumps through hoops trying to show that this is all really a good thing in the long run.

There are other possible explanations, and one is "externalities". That is what happens when costs aren't properly shared, so the people getting the benefit don't match up properly with those paying for it. For instance, in Britain there is a mismatch between costs with road and rail transport. Trucks use roads which have been partly paid for out of all tax revenue, not just from road taxes, but railways are maintained by funds which are well matched up with the railways. That ends up favouring roads over railways, because ordinary taxpayers are subsidising the road transport industry.

To give them their due rational economists know all about externalities, and they know exactly what to do about the problem. Just make the costs go to where the beneficiary is, so everyone knows what's really rational. That's where the slogan "user pays" came from - it's a cure for all the problems caused by externalities. But it's more subtle than that, and the rational economists sometimes miss a couple of things. First, the trick is spotting just who the user is. When public urinals were introduced last century [now last century but one, i.e. the 19th century] the "user" wasn't obvious. It was the householder whose doorstep was no longer being used for the purpose - the person getting the benefit, not the person doing the deed. Second, there are some kinds of mismatch that aren't really externalities with the standard solution. For instance there's something called the "tragedy of the commons", known for nearly two centuries. It's not an externality because its cost mismatches do not come from one group paying for another - there's only one group involved, but in two different capacities.

Here's what happened. In the old days English peasants - commoners - literally had various resources in common and the village green really was a common they could keep animals on. But there is a limit to any common's carrying capacity. If the yield of the common starts dropping off when there are twenty animals on it, so twenty-one yield as much as nineteen and so on, there is a problem. During the seventeenth and eighteenth centuries village populations started reaching the limit - say, ten extended families with two animals each. Each commoner had to decide individually whether to put on more animals. The numbers say, add one animal and total yield drops - but whether the others put on more animals or not, each commoner's share is better with one more than without. The best interest of all can be very different from that of each.

This pattern happens in other places as well, such as the way all customers pay for the operating costs of credit cards whether they use them or not - the cost goes on the price of everything, whether bought with a credit card or not. If you tell the banks about this they patiently tell you that it is all really being paid for by growing the market. But that's missing the point - there is a tragedy of the commons whether the market grows or not. If the market grows enough to keep prices as before, that compensates, but nothing says it must. [Since writing the above, Australian law has changed to allow retailers to charge credit card users differently.]

That's what's wrong with the idea that productivity gains always help us all by creating enough new jobs even though old ones are destroyed - nothing says they must. And it turns out that along with other things there is a tragedy of the commons helping to drive up unemployment. Social security is a cost of unemployment, but it doesn't show in the numbers that employers face - it's part of all employers' tax, not of each employer's tax. No individual employer feels the yearly $10,000 or so social security cost of each retrenchment. Payroll tax makes things even worse.

But we could offset the bias by rebating $10,000 from employers' yearly tax bills for each employee, raising the general level of tax to compensate. Taxes would only look higher, because after the rebates the tax bills would stay the same - and they would drop as unemployment eased. Less tax wouldn't hurt the government, because the social security bill would narrow in step with the tax base. Employee rebates would encourage apprenticeships, discourage retrenchments and bring the long-term unemployed into the workplace by privatising social security. And a GST with these rebates might be progressive enough to be acceptable.

Return to the contents list.

Article printed in News Weekly of 22.8.98

GST not a consumption tax, and some variant tax options

There have been many discussions of a goods and services tax or GST (or value added tax, VAT). Most of these have looked at what could go wrong or - from its advocates - what it should do, either in itself or as part of a package, in both the short and the long term. But that is only looking at the theoretical possibilities. These are limited by some features that are inherent in any GST. That is, no matter what we aim at or how things might change later, there is always an overall pattern.

In particular a GST has two major features:-

Other countries' experience backs up the first statement. The truth of the second statement is not so obvious. Most people - even some experts - think that, since all the tax is passed on by merchants and the consumers eventually pay, that makes it a consumption tax by definition. But that is an illusion, because while the end payers are obvious some intermediate effects are not.

Think what happens when a wholesaler sells an item for $100 pre-tax, the retailer marks it up by 15%, and there is a 10% GST on top. The retailer's numbers go like this:-

Most discussions don't show that second step. GST advocates frequently claim that improved cash flow from holding onto the $1.50 will pay for the compliance costs, but they leave out the hidden cash flow penalty from the tax prepaid on the working capital. It is true that the cash flow is better than with a WST (wholesale sales tax), particularly if there was a provisional aspect to that, but the improvement is only relative - and a later government might put in a provisional tax element. On the whole the retailer has poor cash flow, and the little benefit there is could easily be too small in relation to the compliance costs. How it works out all depends on the proportions, the timing, and how much use the extra management accounting information is to the retailer - which is why the compliance costs are uneven. Some businesses, especially big ones, need that sort of information anyway and there is no particular additional compliance involved, but for others it is a huge additional burden. In the example above, the retailer would need to hang on to the tax for literally years to get any material benefit.

For an economist most of the difference between consumption and production is investment. This includes working capital with value in trade as we have just seen, but it also includes fixed capital with value in use. For instance, if a newsagent buys a car for deliveries, that is fixed capital - it will get used in the business. Now, in some ways it is easier for a GST to adjust for that, because such purchases are less frequent and they can be classified as business inputs to generate a rebate. But is it so easy? First off, the exercise of classifying is not straightforward. There might be two identical cars in a showroom, and one gets bought by a newsagent for business and the other gets bought by a commuter. Before the sale they are both working capital, and afterwards one is fixed capital but the other is a consumer item. You can't tell just by looking at them. Just to complicate matters, the owner of record might be a bank with a charge over fixed assets, or a manufacturer retaining title over goods with a contract with a Romalpa clause. Who gets the adjustment?

Even worse, the classification as business inputs must be acceptable to the government. A few years ago there would have been trouble classifying computer software - and even today you can't depreciate it fast enough. Classifying fixed capital as business inputs is the same thing as the government picking winners - it requires the bureaucracy to make accurate judgments about what is or is not appropriate in running a business.

Such adjustments are possible, but they raise compliance costs and they are necessarily imperfect. It has been suggested that Australian compliance costs can sometimes be as much as the value of the work being done, in this case the tax itself, and - since Australian regulations change frequently - few transitional costs are truly temporary; before one lot of dust settles another lot gets stirred up.

A consumption tax is supposed to encourage savings by affecting consumer prices, and these are the same with a GST as with a true consumption tax. So why does it matter if a GST is a production tax? Because of that effect on investment. There would be an incentive to save - it's just that investing in Australia gets correspondingly harder, which counteracts it. Other things being equal, the net effect is to encourage people to invest abroad if at all.

There are a number of theoretical ways to turn any production tax into a consumption tax. You could allow interest on the rebates on prepaid tax, but it is next to impossible to allow the right interest on rebates to make the adjustment, and it would raise the compliance costs. A simpler approach is to grant interest free overdrafts corresponding to the prepaid tax. But it's still not very simple - it requires all the investment to be categorised, just as before. Compliance costs are like moving the wrinkles in a carpet around - whatever you do to claw back a problem just gives you another one.

There is another direct way, the "People's Bank" option, granting businesses low interest loans. Remember that interest free overdrafts on a proportion of capital would have been equivalent to making the right adjustment? Well, in their turn low interest loans against the whole business are the equivalent of that. The difference is that the loan decision is uncoupled from the tax issue and compliance is easier, because it is no longer necessary to justify the loans against tax-related information (but there are other difficulties with this approach).

As part of a tax reform package, however, none of this matters - provided we remember that we have not got a consumption tax and some other part of the package needs to help investment. One way is for the package to encourage savings that would flow all the way around to investment and make up the difference - Singapore and Chile offer examples we can look at. On the whole this seems the simplest way with the fewest compliance problems.

That still leaves the overall compliance cost problem for the original GST. There is a standard way to reduce compliance costs in the tax area, called "composition". With this the bureaucracy allows a grossing up and perhaps a shift in the payment point, so absolute detail is not required. For instance, while a strict PAYE (pay as you earn) system makes income tax easier for wage earners, the moment you have interest on savings your tax return gets harder. In the U.K. a special concession made life easier for the ordinary tax payer by allowing building society and post office savings to earn tax-free interest, at a lower rate because a standard withholding proportion was paid directly to the government by the institution. Similarly gifts to charities might be matched by government top-ups rather than making the tax payer get a refund on tax. It's necessary to place limits on who can do these things and how much, to prevent abuses - but the gain in simplicity saves people a lot of unhealthy stress.

We can use the same composition approach to simplify a GST - perhaps even completely out of existence. Bearing in mind that we can correct any production tax towards a consumption tax with the other parts of the package, we can turn our attention to other production tax options. We can either address production directly, with some form of turnover tax, or we can do it indirectly by taxing the factors of production - capital and labour, and classically land as well. From an economist's point of view all land is treated as capital these days, but from the taxing point of view it is easier to do things the other way around and focus on commercial plant and premises. Taken together these lead to four kinds of tax:-

The last two kinds of tax can reflect business situations well, provided the businesses are not too diverse. This is because there is already a statistical measure, the Du Pont formula, that brings out the appropriate profit levels for various sectors of the economy. This could be used to vary the rates. A turnover tax can be safely set higher because it adjusts for lower profit levels, but a tax on the use of commercial plant and premises can be applied to a wider range of businesses, even those with a range of activities, as the amounts owing simply add up in a straightforward way. Conversely, the more we have a fixed tax component the less we penalise profitability - economists have studied this in comparing fixed rents with sharecropping.

This means that a tax on the use of commercial plant and premises could displace a GST, but only until it burdened low-profit businesses. Since it can be applied locally it could be used by the States, replacing payroll tax - indeed, since payroll tax harms employment, a negative payroll tax could help it. If producers got a rebate matching social security for everyone on the payroll this would tend to bring people back into the workforce. The overall tax level would appear higher to compensate, but that would be an illusion. Total tax bills would be the same.

A turnover tax would need to have rates that varied with the Du Pont measure. You could only do this with simple businesses with a single rate, so you couldn't do it across the board. But these are exactly the businesses with serious compliance cost problems. We could use a turnover tax as a composition for GST for those businesses meeting certain criteria, say fewer than a cutoff level of employees and only one whole or main category of business activity. These businesses would be unable to concoct false returns plausible enough to escape modern statistical analysis.

To sum up, a GST could only work properly if other parts of the tax reform package encouraged saving with flow-on investment in Australia, and composition with a turnover tax can often be used to reduce compliance costs where these are excessive. State payroll tax that harms employment can be replaced by a tax on the use of commercial plant and premises, with a negative payroll tax component to encourage employment.

[Afterword: over time, a GST can provide some relief for tax on investment by allowing prepaid tax to be offset against tax due on current profits until the accumulated prepayment is eliminated; this resembles the interest-free overdraft approach, except that it is only made available very slowly and the compliance costs are even higher - it really only provides adequate adjustment in those special cases where capital requirements can be funded properly by retained profits.]

Return to the contents list.

Article printed in News Weekly of 17.10.98

Never make an engineering decision for accounting reasons

[This was written before the 1998 Sydney-Hobart race, and the example used may or may not describe the events of that race adequately.]

First New Zealand electricity, then Sydney water, then Victoria's gas - all have experienced major interruptions in the very recent past. Many people reached for their pet theories looking for the reasons, and ended up blaming privatisation, or failing that the corporatisation which often precedes it.

It cannot be as simple as that. Some of these utilities have been privatised for a long time, or haven't yet been privatised or corporatised much; and their equivalents in other countries haven't experienced similar problems. So we have to look deeper.

As various newspapers correctly observed, all these problems have a similar immediate cause: many systems have been cut back to the bone, eliminating slack in the name of efficiency and cost-effectiveness, so when trouble hit there were no reserves. This sounds like downsizing, only applied to infrastructure instead of people, so to that extent the accusations were fair. The problem derives from the same spirit of the age that shows up in privatising, even if it is not privatising as such. It has even happened with vital agriculture. Gibbon supposed that Europe was safe from the massive disruption that Rome got from the barbarian invasions because agriculture had improved and broadened its base; well, we have now reversed this - less than 5% of the U.S. workforce is directly engaged in agriculture, which is highly vulnerable.

How could we let things get that far? More to the point, what stopped it from happening before? We can look deeper and try to find an ultimate cause. We can look for similar cases in other fields, and then see what they have in common. A good one is ocean yacht racing. Every few years, in races across the world from Fastnet to Sydney-Hobart, yachts have been lost from the same immediate and ultimate causes. These are by now well known, but short of a firm hand by the race organisers the yachtsmen drive themselves into the same situation continually, like moths to a flame. What is the immediate cause? The yachts are built for speed not strength and stability, so they are not seaworthy enough for the bad years. What is the ultimate cause? Winning is the only thing that counts - coming second, even by a fraction, makes all the difference, so there is every incentive to keep shaving the margins.

But the trouble is that it's like the old engineer's joke about the right amount to tighten up a nut - just tighten it until you strip the thread, then wind it back half a turn. Pretty obviously that is the right amount to turn it in the first place, equally obviously the catch is that it is too late finding it out this way, and the point of the story is that there is no other straightforward way to work out the right amount. Since what is a safe yacht in one year is not what is safe in another, some designers end up getting away with things rather than learning the limits of what works - and the same applies to the regulating skills of the organisers, since between failures valuable knowledge fades; only bitter experience teaches it again.

Yachting competitions are non-linear - instead of rewarding twice the effort with twice the result, there is a sharp break and cutoff, which encourages cutting corners. We know this mechanism from physics, where it drives "relaxation oscillators". Most economic theory focusses on easier mechanisms which are linear or nearly so. This is not so far wrong - a relaxation oscillator is part of a violin, for example. What happens is that there are enough other things going on that the near-random oscillator mechanism gets smoothed out, and this averaging out makes mere noise musical.

That is what - mostly - happens in economic competition. If several people tender for a contract there is only one winner; but there is always another day and another contract, so it should smooth out, by and large. Only, it doesn't always. One reliable way to go broke is to gain an increasing share of a declining market sector - there are no tomorrows there. Contrariwise, you can get away with cutting a lot of corners in an expanding market - this year's orders are big enough to bury the costs of last year's mistakes. With high inflation this even happens with a steady or slow growing market. In either case the culture of the market sector can develop some very bad habits that cause trouble later when it goes into decline.

That could cause some of the trouble, but it still does not explain why it never used to happen, nor why the averaging out isn't good enough. There are two answers: it did happen, only we have forgotten it and the precautions against it our forefathers learned; and we now have institutions working to exaggerate the destabilising effects. In nineteenth century Britain there were a great many gas explosions, train crashes and so on, so these days, all over the world, trained gasfitters now install gas appliances and trains have safety systems. For an Australian example, there's the Goyder line. Some early settlers thought that rain followed the plough, so they pushed settlement further and further back; but they were ignoring Goyder's survey, which marked out the limit of reliable rainfall. As with the yachtsmen, the good years encouraged them to go too far.

This shows one possible cure, giving people proper guidance like Goyder's; once people took it seriously, they followed it. But modern institutions make that harder. In particular, even when companies know they should maintain a strategic margin, reserve or slack they are penalised for doing so - the way the Adelaide Steamship Company was, because low debt makes takeover targets. In theory this sort of thing maximises shareholder value; but the valuations don't always make that difficult judgment of how close to the line it is safe to go. That is, while corporate finance knows very well that the bottom line should also allow an offset for risk, the rest of the bottom line - from the gross profit - is easier to assess objectively than the risk factor. Working that out is as hard to judge as knowing how much to tighten a nut. With accounting the hard part isn't knowing how to add but what to add - categorising everything, leaving nothing out and not double counting. But since the entrepreneurs who go closest to the line get in first in today's way of doing things, they scoop the pool and end up in control of those companies that play it safer. They win out in the short term and the others don't even make it to the long term, so this enforces short-termism.

In the old phrase, companies make engineering decisions for accounting reasons, which is just as bad as making marketing decisions for engineering (operational) reasons or accounting ones for marketing reasons - while they should be kept in touch with each other the tail should never be allowed to wag the dog. Add in the idea of the spirit of the age, the way government-run bodies draw on the private sector for their best practice, and there you have it.

Return to the contents list.

Article printed in News Weekly of 9.1.99

Australian savings and investment, why and how

Why do we need to save more? At the moment Australia has a number of economic problems - or issues or opportunities, if you prefer. They work together, so even though you can start addressing them in isolation, as with my suggestions in April [1998] on attacking unemployment with a Negative Payroll Tax, you have to look wider.

For instance, any approach to improving employment has side-effects, some all its own but some just because there are more people employed. A happy and prosperous Japanese tends to save more. A happy and prosperous Australian, on the other hand, tends to go out and spend - and much of it goes on imported goods and services, which tends to blow out the balance of trade. Since we already have a problem with the balance of trade and the balance of payments, any way of improving employment would always aggravate this.

In the short term we can do something about this, but at a cost. We can do things to help the balance of payments with primage tariffs (still barely permitted), but that does not fix the underlying problem - the increased government revenue would still need to be channelled into private savings and investment somehow. With taxes on the primary industry content of exports, e.g. taxing their bulk and weight, and a floating dollar, we can get the same effect and encourage Australian industry to add value - but that gives the illusion of a burden on primary industry, which makes a political cost.

So the only real long term solution must involve an improvement in Australian savings, together with channels for them to flow through to constructive investment in Australia. Improved savings would help the balance of payments and inflationary tendencies directly, and would loosen the policy constraints on government spending. Constructive investment would mean that there would be a real payback later on, because when people drew down their savings there would be more opportunities to spend on Australian goods and services rather than on imports.

How to do all this? Well, that's the trick. One focus is on superannuation, either by individuals or by governments acting on their behalf. This partly works, but it faces two kinds of problems. One is that every increase in superannuation measures reduces personal incentives because it reduces personal involvement, and the other is that by interposing middlemen it restricts the channels that direct savings into constructive investment. We have middlemen for prudence, so as to make wise investments - not just to maximise our chances of getting more but to minimise those of losing the lot, partly by spreading the risk over different generations.

In a perfect world everybody would be far enough removed from personal distress that we could all look out for our own futures and none of us would need a safety net. Well, it's not like that, but we can imagine it and see where it takes us. Engineers study theoretical perpetual motion machines, not because they seriously expect to build one but because they know that the things that make them fail show the limits of real machines. In the same way, we can look at an imaginary perfect world and see what we need to do to make up for not being able to reach it.

Suppose we switched overnight to a regime where everyone funded his or her own superannuation. Straight away there would be an injustice and a failure. The injustice would be that people who had paid all their lives for the age pension would suddenly find it wasn't there. But there would be a failure too - nobody still in the workforce would have the right habits to save enough, and they wouldn't even have the means if they were unemployed or still faced the previous levels of income tax.

This gives us a clue. Still in the spirit of a perfect world, suppose that there were no unemployment or other factors that couldn't be eliminated at the same time as age benefits - that the income tax load could be wound back in step with age benefits and that people had genuine opportunities to make their own arrangements (we are talking about a perfect world, remember). Then we could phase out government pensions and compulsory superannuation by the grandfathering method - that is, everyone already retired at a cut-off date would remain beneficiaries, but no new retirees would be allowed into the system. That would be just to people who had already paid their way in earlier life. Only, what way would there be to give back the cost reductions as income tax cuts that would both be just and would allow people to make their own arrangements? On the face of it, none, because older people would not accumulate enough income tax cuts and younger ones might not develop savings habits early enough.

People who just missed out on pensions would really miss out. There is a way to reduce the injustice though - instead of restricting pensions to those who were already receiving them, just raise the eligibility age gradually, say by one year for every two that passed. That way at least those who suddenly had to provide for themselves would only have to save a little to cover a brief period, which is manageable - but it's still unjust.

Let's try again. Suppose that instead of grandfathering out age pensions completely, we just clawed them back a little, so they started a couple of years later. What then? People coming up to the former retirement age would deserve a greater opportunity to save, but would not face an insuperable hurdle - they would only need to build up a small nest egg.

Since we are talking about different phases of life that we all face, the equitable thing would be to concentrate the income tax cuts on these last few years of people's working lives.

It would be more just because the older taxpayers would not miss out on tax cuts later generations could benefit from throughout their lives. One useful thing about this is that people in this older range today happen to have greater savings habits than younger ones, so income tax cuts would translate into the necessary savings for these.

As younger ones entered the intermediate range between tax cuts and deferred pensions they would have the example of their predecessors to encourage them to save in their turn.

This is promising. We now have a scenario in which things are unchanged for older and younger groups, but both pensions and the taxes funding them have been eliminated for the intermediate age range (not taxes funding other things like defence, of course). Can we go back to grandfathering to increase the scope of the reform?

We can. By combining both sets of modifications - raising pensionable age more gradually than the passage of calendar time and lowering the age at which income tax rates drop in a way that matches the actuarial requirements - we could equitably drive a wedge into the present system, leaving an ever-widening window for people to make their own arrangements.

In a perfect world, that is - which this isn't. What's still wrong with it? For one thing, not everybody would have the opportunity to receive tax cuts - a point being made about tax cuts offsetting the impact of a GST. For another, we don't all have constructive channels into which we could pour our savings - for instance Japanese savings very often drive the inflation of asset values rather than increases of productive capacity and actual production.

There are some answers to these - partial answers, but in my view enough to make it workable.

First, we would always need a real safety net - a government pension as it was in the early days, guaranteeing survival. This aims lower than superannuation schemes that aim to provide comfort levels, but it is as low as we can afford as a community, if we can arrange for people to provide comfort for themselves.

Second, people should be allowed a holiday from the self-funding window for every period where it was no use to them. Someone who was unemployed just before retiring wouldn't be able to use that period to save - that person would deserve to have his or her benefit entitlement age brought forward in proportion with the wasted part of the savings window, i.e. for any period where official unemployment overlapped the window.

Third, there would need to be real investment opportunities - this corresponds to the reality underlying discussions of whether schemes should be funded or unfunded.

People could of course invest in private superannuation schemes in the old way, but this is actually an absurd choice for many of us, because the benefits of their management decisions are marginal and the fees are imposed at the beginning - with any real career mobility nobody gets their money's worth.

Further, we are also considering the public benefit of people exercising their judgments over investment. What we need is a real collection of investment opportunities which is reasonably safe and accessible to all.

Actually, there is at least one useful area already - debentures secured over the working or operating assets of businesses. This market is called securitisation; it is not fully developed, but it does exist and it could be widened.

With the need being phased in by a wedge of age ranges opening a window to people who could save and invest more, it is likely that this and other markets would develop fast enough to be constructive rather than just flooding with cash chasing investment opportunities.

However, we do need a precaution: the government needs to be able to call a temporary halt to the widening of the wedge if there were signs of this. But since the funding needs of the existing superannuation schemes would also be changing and there are some foreign opportunities too, things should balance out over time - and the point is, we would be allowing that gradual adjustment.

Where does this leave us? There are some loose ends, mostly to do with the shifting demographics of an ageing population - but they do not concern this article because they are present no matter what we do. They have nothing to with whether Australia needs more savings and investment, nor with whether this scheme could help those - only with whether people's needs might remain unmet. That is a demographic issue common to all retirement options, so it nearly balances out between the options (except for things like migration policy) - but we should address it at the right time and place.

To sum up, we can help Australian savings and investment by:-

Return to the contents list.

Article not yet published

Australian Population Change

I wrote an article for the News Weekly of the 9th of January, 1999 which had a loose end in it. The article was on Australian savings and investment, and it touched on the shifting population base because of how Australia's needs vary with people's ages.

The loose end was "nothing to with whether Australia needs more savings and investment, nor with whether this scheme could help those - only with whether people's needs might remain unmet. That is a demographic issue common to all retirement options, so it nearly balances out between the options (except for things like migration policy) - but we should address it at the right time and place."

This is the time and the place.

Australian population figures connect to a wide range of social issues, so we need to disentangle those connections as much as possible first. Here are some of the connections:-

The first point was mostly covered in the earlier article. What that article didn't cover was those things that are fixed compared with the economic incentives, or else vary so slowly that it is realistic to treat them as fixed (a mathematical technique called "perturbations"). What happens is that when individuals save for their futures, or else when governments make funded pension arrangements, this makes a difference in some areas but not in others. If I invest to cover my future needs for food, this goes - circuitously - into increasing the world's ability to supply food, by breaking new ground or introducing new machinery, say. But when I invest in my future medical needs this can only be met by arranging to have future medical services. Since doctors can't - or shouldn't - be stockpiled, this can only mean arranging either for more future doctors or for diverting some of those future doctors to me. In economic terms this is a question of "real" rather than "nominal". But the doctors of twenty years' time are alive now - as five year olds, maybe, but alive. So this is where the economics argument falls short - we need to consider the demographics as well.

The economic argument doesn't care much about other issues. All that matters to that is that there should be enough all up numbers to meet the future's specialised needs for labour. As far as that goes, it doesn't matter whether future doctors are diverted from being lawyers, whether they are migrants, or indeed whether enough potential patients die young that there are enough doctors for the remaining old people. Economics can be morally neutral and indifferent - if we let it. What counts for the economics is that saving now to provide for future needs just pushes the price up if it is in an area where demographics - people numbers - are what count. Mostly, saving drives investment which both pushes prices up and increases supply, in a relationship which varies with what economists call "elasticity", the way spending now builds nursing homes which will be around in twenty years' time. But since people are not slaves we have to deal with people numbers in other parts of public policy - economics still matters but it is not on top of the requirement.

Since we are talking about ourselves and not just about statistical abstractions, we know that we need a comfortable age distribution to continue in Australia. We also know that the middle-aged for twenty years' time are alive now, so that means we need immigration or to brace ourselves for a tight squeeze - we are technically allowed to choose the latter.

That's basically where the last article left us. Now, suppose we choose the first option - immigration. The checklist above broke down the issues into two. Traditionally countries met their needs from a sort of "internal immigration", a flow from the country to the town. This worked reasonably well because - with the public health and food supplies of earlier centuries - the fertility of country women and the child mortality in towns meant that the towns needed mature people and the country had young people to spare. Malthusian principles maintained a brutal balance, or else land shortages encouraged later marriages when overpopulation loomed - or, in more recent periods, people emigrated completely.

That brings up two points. The first waves of Australian migrants were from a single cultural background, so the only issues were of this sort - we can look at that simpler age without needing to analyse quite so many cultural issues as we do now. And, internal migration is still going on in Australia, from the country to the town and from one state to another - which means that, looking at the effects that spring from those, we can see how those issues work in isolation even under modern conditions. Once we have dealt with that we can survey the broader effects of today's immigration patterns.

When a surge of Victorians move to Queensland this has two major economic effects. There is a shortage of public and other services, so there is a strain on the revenue base. And the increased local demand for private goods and services boosts the local economy, and some capital is drawn there - perhaps away from Victoria - while some savings were brought there along with the migrants. The local boom is illusory to some extent. The increased demand for groceries is permanent (unless the migrants move on again), but the needs for building are a one-off. Unless the initial boom translates into a permanent increase in the size of the economy it subsides again.

The same experience showed in the 19th century migration patterns. Compared to the U.K., food was both more plentiful and cheaper here (though more monotonous), but lodging was worse and dearer - quite simply the buildings weren't there yet, though the land was. That led to the historical land boom and bust, when the need was more than filled (although there were other causes too).

When this sort of thing happens it means "growth" was spurious. Yet, we see that the boom era did leave more at the end than there was at the beginning. What it comes down to - ignoring the difficulties of transition for the moment - is that in the long run more people do add value, even where there are diminishing returns. If Australia were just a quarry it wouldn't help to have more Australians around - the economic rent, the extra that comes from those natural resources, would just be spread thinner. But if those Australians add to what they receive they really multiply it - and if they not only add to it but retain their efforts, invest them in the future, the multiplier becomes exponential growth.

So it was in the old days. But, it's not happening now. First, because of diminishing returns more Australians are not equally productive compared with earlier ages - the value we add can only work on what is there, both free from nature and as a result of our own accumulated investment. There is only so much in the quarry, and we are not investing to make up for what we use up, let alone to add to our existing store against the evil times.

Second, there is an effect of globalisation. Value is no longer being added in Australia - it is being added overseas. To some extent the economic rationalists are right, that if Australia deserved investment Australia would get it. But they omit the role a conscious example plays in becoming economically "worthy" - we don't save and invest, and our attitude to globalisation does not encourage improvement. There are temporary measures that work like training wheels on a bicycle, both allowing progress now and developing good habits to develop for later - yet the economic rationalists do not allow this. They appear to want sentence first, trial later. So instead of taxes on the export of primary products to encourage value to be added before export, and instead of savings incentives and investment opportunities, we get open trade to low labour cost countries that exports employment and a GST that encourages overseas investment.

That means that we could benefit economically from increased migration, particularly if business migrants brought savings we have foolishly foregone - though not as much benefit as in earlier years when there was more natural bounty to share - but it would be futile unless we had measures in place to retain the boost they gave. That comes back to the economic measures I covered in earlier articles.

There is another traditional argument for numbers, namely the defence one. That has not gone away, but it is no longer fashionable. Suffice it to say - although this too needs another article - that if the Australian Defence Forces are low and getting lower, so that today they can only half fill the MCG, there is not much point in having more potential recruits if we are only going to waste them like the potential investment they bring. The only advantages are that there is not much risk of a military coup and the MCG could hold quite a party.

Now we can look at immigration issues themselves. First of all, we now see that the experts have made one true but misleading statement. Migrants do add value to Australia. Only, they don't do it straight away - even without cultural issues there is a transition, and if we grow continually we never quite receive the gains, yet if we stop the sudden slack can bring on a bust. So even without cultural issues there is a downside - but one that can be managed, if only we do things right. The pity is, we aren't doing things right.

The second thing is to wonder why, if we aren't even structured to collect any advantages, there is such a political momentum in favour of immigration - yet such a decline in grass roots support for it. Here we are venturing into the realms of human nature and group and individual psychology, but I can offer two hypotheses which could be tested. Grass roots support declined when the migrant groups stopped being examples of the culture already being represented - whether you call it redneck or not, people weren't feeling a present benefit from more brothers at their backs. Yet, politicians favoured it, as did the media. I have heard it suggested that the media favoured it as a "status seeking activity", that by being different they felt elite, but that can't be the whole story for politicians - surely they would have had to answer to the grass roots? To some extent this is indeed happening, yet there has to be a reason why it took so long and had so little effect. My own view is that representative democracy has "agency costs", just like the way a company has because its managers' interests may differ from those of shareholders. By instalments immigrants change the political base, so what a politician loses in grass roots support from migration - he or she gains from the migrants themselves! The politicians were "electing a new people", whether they realised it or not - and whether we want this or not is certainly a question which should be asked rather than let slide.

This is one of the defects of democracy, that it cannot, does not, define "we the people" because to ask that is too circular. Yet these questions of identity that we are forbidden to ask lie at the very heart of such important questions as the republic debate. We do have one proof that this can happen - ancient Athens, a direct democracy, had an opposite incentive - to keep out valuable migrants so that the smaller cake would give larger though fewer portions when divided. In absolute terms that stopped Athens' growth, just when it needed it for defence reasons - so we can go wrong either way. All we know is that there are structural biasses that obscure rational insights into the immigration question.

Returning to the issues of different cultural values, without even enquiring into whether it would be right or wrong, wise or unwise, to let in multicultural immigrants, we can see two things. One is that there is absolutely no reason why we should make things even worse for our own - we should straight away stop hindering the bush and stop hindering Australian families. And the second is that we should face up to cultural differences and treat them not with a barrier but with a bridge - and recognise the beam in our own eyes too, because if we went further than stopping the harm we do to natural growth, if we tried to encourage it, that too would be a form of social engineering and cultural change.

What bridge could be built for other cultures, and what internal mechanisms could we have to encourage natural growth? My own mother was once a migrant, of an Irish family that emigrated to France immediately after the First World War (I'm sure some readers know exactly what that means). She re-emigrated to England partly because the French were so unaccepting, which is why I'm here now. This is no way for us to go - yet we could easily throw the baby out with the bathwater. Multiculturalism rests on tolerance, yet the world is full of intolerant cultures. We must provide for education at that level, both directly and by allowing in people at those stages of life where they are more adaptable - teenagers are just as good an investment as breadwinners. Oddly, it is the isolated mothers of families who don't make that contact - so we should encourage singles over families, and let them plug into the mechanisms for natural growth that help Australian families.

Other times and places found partial solutions to the problems of a population shortfall. The French prevented shrinkage with land laws that preserved family farms - yet they prevented any further development, and caused stagnation. The ancient Romans found that large households with slaves caused growth - but of lots of little slaves, not lots of little Romans, so they were changing away from the very population base they wanted. Need I say, both those population needs were for defence reasons? We really must revisit that issue.

But most lasting of all, the Byzantines had a tax system - the Kapnikon or hearth tax - that encouraged large, supportive families because those paid the same tax as small ones. What happened was that they needed a tax base, and this didn't narrow it, just concentrated it - it's the same "tragedy of the commons" mechanism I described in April 1998. This actually worked, except that the reason it was needed in the first place was a series of separate drains on the population, and the Byzantines never fixed those - so the supportive families weren't able to make enough difference. However it worked through to the present day since it was continued under the Ottomans, and that led to the present majority of Albanians in Kossovo [sic] - the "revenge of the cradle"; that really was Serbian once, by any reckoning.

That tells us how - long term - we can fix the problem of small, unsupportive households. Just revise personal taxation so it falls on households, rather than as individual income tax - and grant discounts for dependents of all ages whether related or not, so au pairs and early retirees could help within the families. The price is that we may not all want to live in extended families, and that it takes generations to cut in.

Is there another way? Yes, we can copy Caroline Chisholm, who started by helping a group of non-English-speaking migrants (Highland Scots who had only the Gaelic - which she knew - and for whom she found work as contract woodcutters). Her later work involved placing young girls in what were effectively au pair positions with young families in the bush. This both supported the families and introduced a new generation to the bush in a responsible and secure way - because they could and did go on to marry, since there was a surplus of young men there in the conditions of the time.

So we can - and should - encourage employment in the bush, together with au pair positions in families, all met either by willing Australians or by young migrants on short term visas with an option for staying on if married - let's say, if they became parents, as with the old institution of handfasting. That way it would not be a prejudice against migrants but would provide exactly the bridge they would need to experience Australian values and decide for themselves if they wanted to take them on board - and it would not be a barrier, as it would in no way prevent other migrants arriving in other categories. But I see no harm in allowing any young, willing and non- destructive people to come, for to take any other view is to bar another generation from the opportunities we and our ancestors had - and I certainly see no reason why things should work out adverse to Anglo-Celts such as myself. First of all, though, we need to clear out the cobwebs that make any level of migration economically draining, no matter where the migrants come from.

Return to the contents list.

Article appearing here in Spectacle of September 2001

The Best of All Possible Worlds, or The Only Game in Town?

In "The Market Shall Set You Free", in the Spectator Magazine (U.K.) for the 24th of June, 2000, as a preview of a book they were about to release, John Micklethwait and Adrian Wooldridge made a number of minor yet cumulatively effective attacks on [what some people consider are] some perceived weaknesses of economic globalisation. It is worth looking into those areas a little more deeply, if only to see the true costs and harms that go with its putative benefits.

Since they cited an earlier age of globalisation, beginning around the repeal of the Corn Laws in 1846, it's worth looking at some of what happened then. In fact, parts of it were even earlier; the general peace and freedom of trade within the British Isles had already helped Scotland. For, make no mistake, the arrival of sheep in the Highlands, prompted by English markets, led to an undoubted increase in Scottish wealth; only, the Highlands were cleared. This is a type of what is happening now in developing countries, so we cannot just dismiss it as long past or a mere aberration - it stands for all.

Then, English markets made raising sheep for wool comparatively advantageous in Scotland, rather than raising food for crofters who paid what little rent they could. So - and accompanied by a change in how land was held - the sheep drove out the men, who had trouble reskilling to grow wool. While Scotland gained, more than 100% of the gain went to the newly landed aristocracy; there aren't that many percents, but the rest came from the broken men, worse off with wealth than without.

Now, the parallel is a developing country in which subsistence farmers are driven off land and a cash crop grown for export instead - often coffee, which is at present labour intensive. But fewer workers are needed, and the "iron law of wages", a race to the bottom, still applies (particularly since there is often some subsistence land left, so the bottom is only a top up wage even lower than subsistence). Those workers have to take as little as possible, as there is always another unskilled worker even more desperate for that top up wage. But how is that, when our beautiful and elegant abstraction, comparative advantage, has shown that all must be for the best in this, the best of all possible economic worlds?

The usual answer is that this is because the general increase in wealth is accompanied by a wholly unrelated transfer of wealth, and it is as wrong to blame globalisation for the coincidental ills of kleptocracy in modern developing countries as it was to blame the English for the encroachments and clearances carried out by Scots. But this is not so, in these respects:-

Now it is quite wrong to criticise the banks as such; they are mere catalysts and facilitators, quite neutral in all this. But by the same token they catalyse and facilitate any harm that is going, buying and selling and asking no questions. The modern view - which is quite accurate as far as it goes, though one size most definitely does not fit all - is that "debt is good", for with debt you should increase your revenue more than enough to pay the interest; being wise, if this is not the case, you do not do it. But not all are wise, any more than all are good, and what is more, if one borrows, can another afford not to? Sometimes - especially with classic externalities like the "Tragedy of the Commons" - there is little choice. If a farmer improves his yield with fertiliser he must borrow to buy, he does so or the next man does; yet if all do the price drops, since only so much food is needed - more than 100% of the gain goes from the country to the town which gets its food cheaper, and the interest must still be paid. It was partly from such as this that Scotland squeezed out the capital that made Fleming's merchant bank, channelled through the merchants of Dundee.

It may be said that this merely accelerates things rather than aggravates them, that the catalytic effect not only speeds up the harm, it also speeds up coming out the other side. But this is not so. In their article Micklethwait and Wooldridge remark "...many people feel that they just want a bit of a pause. The world has speeded up too fast - even for the winners." While true, this misleads by suggesting that this is a mere psychological effect. But it so happens that speed matters in substantive ways too, because it makes ideas of equilibrium and the long run meaningless - if new change is arriving faster than old change can be assimilated, there is a qualitative difference. Look at the Highland Clearances again. From an Olympian height we could say, nobody owed the Scots a crofter lifestyle; let them emigrate, go to the cities and the factories, or to Canada or Australia. Well, they did these things and we their descendants indeed share in the gains.

Only they didn't. In the short term, you can starve. We know the cities were no carrot to offset the stick, because we have a natural control experiment; on Lewis Lord Lever built Leverburgh around fish processing, and the locals, having more choice, stayed away in droves. And we can see why the cities lacked appeal, when as late as the 1930s a Dundee tenement could have stairways dimly lit by the stairhead gas, flickering from a fitting bent upside down by desperate men to bubble monoxide through milk until it went blue for a cheap drunk, the same diseased cow juice that could give a child TB that rotted through the side of his throat until a hospital could be found. Ah, but at least there was the money for those.

But there were Canada and Australia. Many drowned on the way, in coffin ships, and even arriving brought no relief. Here in Australia Caroline Chisholm found and helped starving Scots who had only the Gaelic and so could get no work; she interpreted for them and arranged work as contract woodcutters. You see, as well as distance there was a cultural journey to make. It did not matter that there were no legal barriers, even so there were effective barriers. Which is enough to destroy any argument that freedom of movement is a remaining barrier that we must take down, to get the full benefit of globalisation; what could Caroline Chisholm have done for human waves of migrants? In physics such things are a shock wave, and our constant change is throwing all this at us, accelerated by our very efficiency. In the 1846 era of the repeal of the Corn Laws, Disraeli was inspired to write "Sybil", and there he wishes for some way to offset the harm to those who are thrown down by change, even as it lifts others up. Even then, those lifted up were the younger generation, the old being dropped.

There's more. If we may not bring Mahomet to the mountain, at least our modern mountains may go to him. That is, we can export jobs. Granted, we in the developed world have a technological edge of sorts, but it is not sustainable. Already some kinds of software are outsourced to places like India. Now recall what I noted above, about how a race to the bottom for wages can go below the cost of living, if only there is some subsistence land about - it gives what amounts to a concealed non-cash subsidy. (Disraeli also wrote of the possible desirability of "potato grounds", which survived into the allotment movement and was an example of just such a non-cash subsistence subsidy.) Although this only happens at the bottom, in any country it works through the local price structure to give comparably lower rates than we can offer throughout. Unless, of course, we do something equivalent but overt, and such we are forbidden. Micklethwait and Wooldridge also wrote "...the far greater gains (the cheaper steel that goes into all our cars and houses) are diffuse and hard to spot." Only, it isn't "all" - it's only so for those that have them, not for the dispossessed, the broken men. With some jobs leaking out overseas at every level, there are always some that do not get a gain.

It is not a question of whether the fact that some might get a greater proportion of the gain might be inequitable. It is the fact that there may well be a paradoxical reaction, the way giving oxygen can make a patient go blind by shifting oxygen away from the eyes. Some may actually go back to make up the more than 100% elsewhere - something that is less likely with slow change, uncatalysed by the engines of finance. We cannot know in general, of course, only case by case; but we have enough of a sound theoretical framework to know we cannot rule it out a priori, and enough anecdotal evidence to suspect it may be happening as we speak, so we should in all prudence cease our rush and examine our future, case by case.

We can look beyond our bellies. Micklethwait and Wooldridge also cover cultural matters. Well, here we are on shakier ground still: are we just burning down our house to roast our pig? Is our loss of our immediate means of support justified by some noble dream of greater cultural wealth - for it is a dream, anchored in the days to come, not here present. But suppose it so, for such things must always lie ahead before they can be sought for. It may also be that these cultural things are our birthright, this cost the pottage we would be fools to keep.

We would still be cheating ourselves, two ways. First, we have already had the cultural enrichment; each new Big Mac is just another of the same, like the old joke that a certain man didn't have twenty years' experience, just one year's experience twenty times over. Second, there's a crowding out. Here in Australia it's reaching the point that multicultural means any culture but our own, no more cooked vegetables but only half cooked or raw (al dente or salads, to you) - and, far from increasing choice, it's more like Henry Ford's "any colour they like so long as it's black" or the school food approach where rather than being offered curry there is always some day in the week when curry is compulsory.

In the end a better metaphor may be, not would we give up our pottage for our birthright, but "what profiteth it a man if he gain the whole world, if he lose his own soul?" Is this cultural bird in the bush at the price of the one in our hand, the things that make us what we are? While Sparta may have shunned the new to avoid compromising the old, even Athens embraced additions of culture without abandoning its own. In this area they only differed as to means. The alternative to multiculturalism is not narrowness but synthesis, a true diversity not self-abnegation.

Return to the contents list.

Article printed in News Weekly of 20.10.01 under the title "Australia must protect the innocent victims of war"

Strategic issues arising from the World Trade Centre Attack - An Australian Perspective

(This was materially edited in News Weekly, and has been placed here with less alteration in Spectacle. The title was changed as shown above - I do not believe there is any positive duty to protect the innocent; it is an act of charity - a great deal was cut and some words, phrases and formatting were altered, sometimes to bridge the cuts. As some of the cut material was significant, this version shows the original draft with most of the new formatting but with the cuts left in. They are roughly indicated by being coloured red.)

After the attacks on Washington and New York and the developments of the past few weeks there are things we should look out for, in case what is happening forms part of some larger scheme, not yet fully clear to us.

Napoleon tried to structure his battles and campaigns using elements from a basic toolkit: engage, dislocate, penetrate, pursue. That's not a straightforward cookbook recipe, just a metaphor for skilled practitioners to adapt. For instance, in dirty fighting the sucker punch works out with engage/dislocate translating into "you hold him" and penetrate/pursue into "while I hit him".

Perhaps the initial attacks were merely "engage" and this US response "dislocate". From the beginning there were real concerns that Pakistan could end up destabilised by US counter-measures, and it is absolutely certain that even if Afghanistan wasn't the enemy of the USA it now is.

Could the terrorist plan be for some real and lasting defeat on the USA? Easily. "Dislocate" places the victim in a dilemma, e.g. needing to be in two places at once or simultaneously moving and staying still. This is what air power delivers; it cannot defeat an enemy, contrary to common belief. What it can do is make it impossible for him to move so other things can be done to him, maybe by the poor bloody infantry.

What happened in Serbia the other year does not disprove this - on available evidence Serbia's will gave when it became clear that ground attacks were about to start, perhaps from the Russians. But there is never a guarantee that the enemy's will will fail. Tamim Ansary has commented on how well placed Afghans are to endure whatever can be thrown at them.

This highlights a common error in understanding military matters. It is not true that wars are just aimed at influencing others' behaviour - there are other possibilities. In Israel the ultimate objective is a secure Israel, and Palestinian acquiescence is not the goal - it would just make it cheaper. But the Israelis are determined to get it whether the Palestinians cooperate or not. The recent attacks on the USA may well have other ends than influencing US behaviour.

So the strikes on Afghanistan can't be the whole of what has to be done. This is particularly true since they cannot possibly be "surgical", even if each and every one only hits real terrorists - the whole country will be disrupted, with some harvests not gathered in, other things not distributed, and so on.

The sequence war, famine, plague and death refers to this - war disrupts, then famine follows, and after a weakened population becomes prone to epidemics which are also more lethal - death comes as no respecter of the innocent.

The attacks on the USA didn't cause significant damage, so many commentators are saying they failed. They certainly got things moving, so now the USA may be unstable and off balance to resist attacks of a different nature, e.g. rocket-assisted mortars fifty miles from continental US oil refining or ordinary sea mines and suicide attacks by boats in the shallow waters of the Middle East. A few months ago there would have been more margin of error, more slack; there are simple counter-measures - but expensive in men, time and money.

That is what dislocation delivers, forcing counter-measures. US forces are getting committed overseas and heavily dependent on a long support "tail" - it is not widely appreciated how very risky an advance on Baghdad in the Gulf War would have been (do you know what happens to an unconverted Abrams tank in a dust storm? what its loss in performance is when protected against that? how very poor it is at outpacing its supplies even with its best performance?).

Make no mistake, the USA is still vulnerable. If Byzantium had not been otherwise occupied the Ottomans would not have got established. This is the significance of "penetrate". At its worst it could leave the USA with an open wound that will not heal, like the Ottomans - "pursue". Or the enemies of the USA may be aiming to use the dislocation to penetrate and pursue somewhere else in the world, somewhere they can reach with the USA otherwise occupied.

Napoleon remarked that with plans "l'execution c'est tout", the trick is doing it. The same goes for our enemies' plans. For us and for the USA, one main defence must be to brace ourselves and not get overbalanced, escaping the dislocation part of the Napoleonic pattern. The USA seems to be avoiding this trap at the moment. There is a further and particular risk for Australia. The right way to inflict terror, Schrecklichkeit, is to pick a suitable victim and annihilate him comprehensively. The victim has nothing to negotiate. What the aggressor wants, and gets, is for some of the observers to hang back or change sides - to split a coalition. It is their behaviour that matters, when they see that the ally of the main player gets no support. We could end up being the object lesson, hung out to dry to show that the USA can't even protect its friends.

We can take positive steps. Americans are not highly thought of outside the USA. Even here we often look on them as benevolent but clumsy, as likely to hurt us by well meaning mistake as anything else, like Lenny in "Of Mice and Men" - and some countries consider the USA outright hostile. So we can play good cop to the US bad cop if the world has cast the USA that way already. The difficulty is, the USA even now will not accept that reading - they often literally do not appreciate what they have done and become. So we should go ahead with good cop stuff independently of US war efforts, even despite US objections. (This is roughly what the British are doing in Oman, which matters for the staging post of Masirah.)

We should establish refugee and relief staging areas, e.g. behind the port of Gwadar in Pakistani Baluchistan, with Australian consular assistance but most definitely not disrupting Pakistani sovereignty.

To work, it must be genuinely and freely offered and accepted, and we should understand how it works too. Good cop/bad cop is so much a cliche by now, few people realise there is a trick to it. What actually happens is, in the face of hostility the subject reaches out for support. That is, the good cop does not initiate his side of things, the subject does.

In a word, we should restabilise Pakistan. That is, repair what the US has just done there, which has the added virtue of heading off our own refugee problems and any Indian encroachment in our area - we don't want the Indian Ocean to be an Indian lake one day. Assistance with local politics works out, since Pakistan would get some patronage over who emigrated. Australia would soften the flow of migrants at source, maintaining a staging area where they could be checked without imposing risks for genuine refugees - this would have been valuable even a few months ago.

This prepares for two strategic benefits. It moves people out of the difficult terrain of Afghanistan into more manageable Baluchistan. And we can start offering genuine intelligence support on the ground. It has been correctly observed that westerners can't be fully fluent in local languages, and we can't pass for locals either. But those aren't the tricks. The first trick is, to understand enough of the local languages - and have that known - that we can't be fooled, since we don't let on just which bits we understand. The second trick is, when going out in search for information, don't try to pass for a local but for some quite other kind of foreigner, perhaps a Baluchi gone to help in Afghanistan.

But, contrary to recent reports, we are not on the same side as the USA, in a larger sense. Indeed, we could with profit repeat de Valera's 1945 speech about small countries caught up in larger struggles. Though we should not be US clients we should be on the side of justice - without seeking to enter into any position of dependence ourselves. In Terry Pratchett's words, "We're just on two different sides that happen to be side by side." If the USA could erode the British Empire to achieve hegemony after 1945 - I see nothing wrong with our eroding that hegemony in our turn, so long as we also build up our ability to survive free and independent. When Americans say we always ask for their help, they are actually referring to this very dependence that they manufactured. They deserve no gratitude for defending us from things they made us helpless from - but as fellow human beings they deserve our every help in their hour of need, whatever we may retrieve from their wreck.

Taken all in all, there are things we can do to help others, to help ourselves, and to prevent fuelling future fires the way western clumsiness has so obviously done in the past. There is no contradiction between bringing justice to bear and cutting back future risks - reducing risks is not blaming the victim.

Return to the contents list.

Article not yet published

One of several mechanisms contributing to modern Australian unemployment, and a way of alleviating it

It is not enough to say, like Calvin Coolidge, that "When more and more people are thrown out of work, unemployment results." That is only an instantaneous effect, frictional unemployment - people who literally are between jobs. The economy responds by moving towards a new equilibrium, so the main effect may be different; in conditions of effectively full employment those retrenched are rapidly taken up elsewhere. In these conditions - almost by definition - not only does the equilibrium involve full employment, but also deviations from it are rapidly corrected. But simple observation of present conditions or a knowledge of historical ones shows that full employment is not the only possible outcome. Not all equilibrium arguments are valid, even if there really is an equilibrium. The delay until getting near enough to equilibrium may be significant in itself; other factors pushing away from equilibrium may have enough time and opportunity to work; or there may be side-effects of being out of equilibrium which have material and/or cumulative consequences.

One approach to improving employment is to eliminate the barriers to reaching equilibrium fast enough - some people believe this can be done by deregulating the labour market. But, while useful, this may be like fixing a leaky boat by improving the pump - the leaks are ignored and remain, there is the cost of buying and running the pump, and it is no good if the leaks still gain on the pump. One worrying possibility is that increases in the general level of productivity might not only help clear unemployment by creating new areas of work, they might also increase the efficiency of the leaks, so to speak. If the leaks are allowed to keep worsening, or the agitation of the pump aggravates the leaks, there may be an ultimately hopeless race between improving the pump and the worsening of the leaks. Something must have raised unemployment in the first place, and we should pay some attention to that. The following is a discussion of one possible mechanism.

Usually any producer planning future use of plant, equipment, stock, etc. is faced with the full cost implications of the choice. Sometimes, however, costs fall elsewhere creating a market imperfection, effectively a hidden cross-subsidy. This happens with employment decisions, since without slavery workers are not assets. When an employer retrenches there is a cost that is not experienced by the individual employer but is borne by the community generally - the cost of social security for each person retrenched plus the cost of its administration. In countries without effective social security there is an even more hidden general cost - the cost of dealing with vagrancy and consequent economic disruption, e.g. rising crime. Ultimately, as everything in the economy is connected, this cost does fall on employers, both directly through the increased tax burden and indirectly through lower levels of demand from the unemployed. Each employer reasons, quite correctly, that all the other employers will bear practically all the burden of those retrenched. Every other employer reasons the same way, but in the end they all pay much the same price as though they each paid for their own retrenchments - there is a separation between the economic incentives and their effects. And something similar happens in reverse when the employer is considering hiring.

Even in conditions of effectively full employment the incentives on individual employers do not accurately reflect the true costs. In conditions of less than full employment not only do the incentives to retrench exist, but they may well help drive up unemployment - because in those conditions there is no necessary and immediate restoration of an equilibrium involving full employment. What would happen if we applied offsets so that individual employers were faced with the true costs of their employment decisions? That would be like fixing one of the leaks in a boat.

Consider those taxes falling mainly on producers, as a GST or a tax on the use of commercial plant or premises would. We can design a system of rebates that matches the costs that employers currently pass on. The size of the total pool of potential workers is about 10.8 million in Australia, with around 10 million actually employed [by optimistic estimates - some measures only have 7 million]. The yearly cost of social security and the on-costs of administration is of the order of $10,000 per head. Suppose we adjusted the tax rates to bring in a further $10,000 x 10 million per annum, but rebated $10,000 per full time employee per annum (net after payroll tax etc., and pro rata for part timers, but not rebating for overtime). The rebates would apply to all employees, new or old, quite without regard to whether they had been long-term unemployed, briefly unemployed or had been employed for a long time - and the rebates would apply indefinitely, with no cut-off. As these taxes would fall on the potential employers, they would be the ones to benefit from the rebates - this makes sense, as they are the ones who would need the incentives to hire and not retrench. The total tax yield would not immediately change with the introduction of this system, but the marginal cost structure would - the cost to employers of each additional worker. The marginal cost of hiring or retrenching to the individual employer would exactly reflect the marginal cost of social security to the community - the employer would face the true marginal costs.

Some likely consequences would follow from these rebates:-

Without labour market deregulation and/or economic growth this would just privatise social security for the otherwise unemployable. This suggests that notional full employment would be possible, even without labour market deregulation etc. - the effects just described would be equally realistic either way. That does not mean there would be no place for labour market deregulation - after all non-jobs would be only slightly better than social security. To return to the boat analogy, as well as patching one leak we still need to improve the pump - the non-jobs correspond to accumulated water that got in before the patch, and besides there might be yet other leaks. While we might have a pious faith that nobody is genuinely unemployable, it would be far better to have a practical faith and do something in the way of retraining, remotivating and such. This system of rebates would undoubtedly help morale and provide a bridge back to fully involving forms of employment, but there is still a place for raising the general level of economic activity, a place for eliminating barriers to finding new lines of work, and a place for providing real training. Providing tax rebates on employees, stimulating the economy, deregulating the labour markets, and providing training and positive motivation are not exclusive alternatives but complementary options. It is not a case of "instead of" but "as well as".

What would happen to the non-jobs? There are two possibilities. The most obvious is that people would gradually move into new areas as they opened up, only now from positions in the work force. The other is less obvious, because we are used to thinking of progress as linear - the new and improved displacing the old and obsolete. But it is more complicated than that. In the early days of the chemical industry chlorine was produced indirectly from hydrogen chloride. Later chlorine was produced more directly, and whenever people wanted hydrogen chloride they made it from chlorine. Later still the plastics industry started producing hydrogen chloride as a cheap by-product, so the old process again became economic - but now the processes complement each other. If the chemical engineers had been as sophisticated as economists they would have known the old one was obsolete and they would never have reverted to it; but they weren't and they did - with complete success. In the same way the availability of people in non-jobs would make them out-compete some of the current methods. As equipment gradually wore out and needed to be replaced producers would sometimes revert to more labour-intensive methods instead. Without high growth new areas could not take up all the slack, but with both kinds of area open it is likely that non-jobs would clear.

This system of rebates would create a minimum wage structure and improve employment at the same time - these goals would no longer conflict. We would actually be creating an invariant, a nominal minimum wage, as all hiring and retrenching would automatically maintain income via either employment or social security. In this respect it is materially different from a negative income tax system - that could only work indirectly by promoting a new higher employment and lower wage equilibrium which would take a while to reach, it would not fund the on-costs of employment such as insurance and increased management costs so it would not absorb the barely employable, it would have its own administrative problems, and since other factors might interfere with reaching equilibrium it might not succeed in full. This makes it subject to the problems of reaching an equilibrium described above.

Provided this minimum wage level was not arbitrarily set too high there would be no additional cost - we are already committed to supplying the needs of survival. There is an identity: while increased employment would narrow the tax base, it would identically reduce the burden on social security - so it is a non-problem for government planning except to the extent that it "hypothecates" revenue, meaning it would reduce the government's discretion to switch social security expenditure back and forth to other areas.

This invariant is only nominal, that is, defined in terms of the paper value of money, not of what it can buy. The real minimum wage would be subject to inflation. As such, and since production taxes do not fall on individuals as voters, we could expect a fall in the political pressures for inflation - although probably not enough to stop inflation completely. Also, fiscal drag - the tendency for inflation to affect the real value of taxes without needing a change to nominal rates - would lower the value of the rebates, so the system would be largely self-regulating; we would not need to get it exactly right first time.

As there would be an effective minimum wage the income tax system could be simplified; as the production tax would already be progressive income tax could be changed to either a flat rate or a surtax only affecting the rich.

Taking all these things into account, it may become realistic to apply some well known but usually impractical tactics of economic management. In particular it is theoretically possible to provide a non-inflationary economic stimulus with the right combination of high taxation and high government spending. With no personal impact of production taxes, with a political pressure in favour of deflation, and with improved consumer confidence deriving from improved employee confidence, this tactic might well become practical.

On the face of it employee rebates on production taxes would help employment and minimum wages directly, would streamline social services, would clear the way for other tax reform and would allow a non-inflationary economic stimulus to be applied - all without obstructing other positive measures such as retraining for modern needs. But we might be missing something - after all, we have not even considered the balance of payments implications. This whole area deserves to be explored in depth, especially looking out for any overlooked aspects, with a view to developing and implementing effective policies and practices.

Return to the contents list.

Article not yet published

A free lunch - at point of sale

By now we've nearly all heard that there's no such thing as a free lunch, but the reason that needs to be said, again and again, is because there is often such a thing as a free lunch - at point of sale. And there are always snake oil merchants who will try to lead you to a lunch that isn't free after all.

Anybody can tell that free nuts at a bar are being paid for by the drinks sales, just as anybody can diagnose a pregnancy once the baby arrives, or a terminal disease once the patient dies of it. The useful part, the hard ask, is recognising the situation early enough to do something constructive with the information - and that's why they have all sorts of specialist economists around, to spot the squalls early and head them off with advanced rain magic. In this case, spotting the sort of voodoo economics that claims to have a free lunch.

Economics is not quite magic but it's not really a science either - there are large parts of craft and art in with the science in economics. A science is when there is a body of principle with predictive power that can be built and tested in a systematic way, e.g. with the scientific method using experiments or some variant thereof. A craft is a body of technique and accumulated knowledge with very little linking one part to another; it's robust because it lacks interdependency, but it's burdensome because when you know one part you just know that one part - all the difference between Chinese writing and a regular alphabet. An art is when some insight is given effect, but there is no standard rule, system or body of technique that leads to it - someone just has to see the insight then do the hard yards working it through.

Economics is mostly a craft, but it is turbocharged by being backed by mathematics, which is an art rather than a science - still recognised by some of the older English universities, which is why my mathematics degree is an Arts one. Turbocharging like this means that individual results can be extended to a wider area than their first discovery, which makes something of a system - but that's a Clayton's science, the kind of science you have when you're not having science. So it is often constructive to fossick around in old workings to see if you can pick up something someone else missed or bring together a connection that wasn't made before.

While you can't get something for nothing, there are two things you can do. You can get something for nothing down - someone else pays, or you pay elsewhere, free at point of sale. And you can get nothing for something. When you undo that - changing it to something for something or nothing for nothing - it looks like something for nothing, but it isn't really.

You might think that nobody would be stupid enough to take nothing for something, but there are special situations where this happens. Let's use mathematics and apply a little games theory to it - the prisoners' dilemma, which you may have heard of.

Suppose a crime has been committed and the police have rounded up two of the usual suspects. They are held separately, and the police make the same proposition to each of them: turn Queen's Evidence and confess, implicating the other man, and you will get off and he will get five years; hold out and either he will confess so you get five years, or we will drum something up against you so you get six months each. The prisoner has been through something similar before and knows it is no idle threat, but asks, what if we both confess? Then you each get two years, comes the reply. Paradox: though each would be better off with six months, the best choice for each is to confess, so each ends up with two years.

There are two intellectually honest reasons you might think that was wrong. One is that you are sure things don't work out that way because the numbers don't add up - like the way they don't add up with a free lunch. The other is you might be sure that the situation can never arise in real life, or if so only momentarily until things work through to eliminate them, the way waves subside at sea.

Unfortunately that isn't true. The numbers do work out, and the situation can arise and persist in real life, which shows up in another real case, the tragedy of the commons.

Here's how that worked out. In the old days English peasants - commoners - literally had various resources in common and the village green really was a common they could keep animals on. But there is a limit to any common's carrying capacity. If the yield of the common starts dropping off when there are twenty animals on it, so twenty-one yield as much as nineteen and so on, there is a problem. During the 17th and 18th centuries village populations started reaching the limit - say, ten extended families with two animals each. Each commoner had to decide individually whether to put on more animals. The numbers say, add one animal and total yield drops - but whether the others put on more animals or not, each commoner's share is better with one more than without. The best interest of all can be very different from that of each.

Now, that is also a made-up example for purposes of illustration, to the extent that the particular numbers were chosen arbitrarily. However the situation did in fact exist, and it was identified in real life as early as the early 19th century - only a little while after it started, and in an era when its associated social problems were very real and present. What all this means is that we now know this is a real and well-understood case, where we can end up getting nothing for something.

What we don't always know, what takes a bit of art, is when something like that is happening in a different context. So let's put that on one side for a moment and go and look at a different piece of theory that fits in at various points of economics - but don't go away, the strands come together later. We are using some of the craft, the unsystematic body of knowledge.

In physics, one of the peaks of late 19th century work - perhaps the last before relativity and quantum physics - was van der Waal's work on gasses. He tuned and he tweaked, he put in fudge factors to make the graphs work and then he went looking for reasons for the behaviour so that they weren't fudged any more, and finally he got something that worked exactly, within the limits of experimental error.

Except for one thing. Over a huge promontory of the graph, the curves didn't fit at all! In the end the reason turned up, and it explained phase changes - how water can move to steam or ice, and so on. It was happening exactly where the curves on the graph reversed their slope from negative to positive, and what was happening wasn't a failure of the theory, just a failure to understand an aggregating mechanism, i.e. how the behaviour of small "pieces" of gas combined to give the behaviour of a larger volume of gas. Catastrophe theory provides a more general aggregating mechanism.

At school there was a simple experiment to illustrate this. You hooked up one gas supply with tubing to two Bunsen burners with the air sleeves closed and a soap film across the ends, then gradually blew bubbles. At first the two bubbles grew in step, which is just what you'd expect, but as they passed the hemisphere shape one would start shrinking while the other kept growing. The thing is that the pressure in a bubble isn't directly related to its size but to its curvature, so it only increases to half way; after that it drops - the slope reverses. That means that after half way if one bubble is even slightly larger it lets in the gas at higher pressure from the other bubble, while before half way the smaller bubble accepted gas preferentially so they stayed in step.

That's what was happening with van der Waal's graphs. They weren't directly predicting experiments with volumes of gas, only the individual pieces that collected together. Where the slope was negative just adding up their values worked, because discrepancies worked to eliminate themselves. But where the graphs went positive addition was the wrong way of combining the results.

This insight carried over into many other areas of physics, like treating the behaviour of magnets with the same theory that worked for phase change in gasses, or handling the change from subsonic to supersonic gas flow with a similar technique - the slope of a graph is changing sign there too.

Getting back to economics, many of these same pieces of underlying mathematics were transferred across to similar problem areas there. For instance there is a close parallel between the change of behaviour of subsonic gas flow to supersonic and the difference between ordinary inflation and hyperinflation - in each case the behaviour when the slope reverses means that downstream disturbances can't signal upstream fast enough against the flow, so any effective feedback in the original flow gets broken.

There are other ways to use the phase change insight in economics. One is to go looking for paradoxical behaviour, where two different things are going on at once, then try to see if there is some underlying behaviour with a curve that describes it that is reversing. If so, there may be a phase change and there may be something we should do about it.

We do have an area where this may be happening. There is some empirical evidence that not only is the NAIRU, the non-accelerating inflation or "natural" rate of unemployment, rising around the world but also those people still in work are working harder - i.e. there is a bifurcation, a splitting into two simultaneous but different forms of behaviour. This rings alarm bells, because it is just the sort of thing one would expect with a phase change (and it's linked to Chaos theory too). The evidence is different in different countries and at different times, but that is entirely consistent with being on different parts of the graphs and moving from contour to contour over time - "shifts". (We deliberately plot the graphs so changes on slower timescales correspond to shifts from contour to contour, while individual curves correspond to rapid or easy changes - the most likely behaviours.) Unfortunately that destroys the predictive power, since so far the theory fits all scenarios, like Marxian economics, and we have to move beyond to get anything useful. Knowing what we now know about phase changes we look for some mechanism that could explain this - something that could cause a slope reversal - and then we look for separate evidence one way or the other as to whether this is really going on.

That brings us back to the tragedy of the commons. That provides a mechanism that reverses the ordinary direction of supply and demand in special cases, so it is worth looking into. Is there a place where it might apply in the employment context? Well, a similar phenomenon is called "externalities", and we know how to look for those. It turns out we can apply a reverse tragedy of the commons to the hiring and firing decision with a Negative Payroll Tax, then perform a thought experiment to see when it reduces the externalities to a minimum.

Let's look at an Australian example. Social Security is a cost of unemployment, but it doesn't show in the numbers that employers face - it's part of all employers' tax, not of each employer's tax. No individual employer feels the yearly A$10,000 or so Social Security cost of each retrenchment. This means that - unless they make the mistake Pullman did before the Great Pullman Strike, of having workers' money flow back to the company - retrenchment diversifies the employers' costs. Someone pays that Social Security, but it's not the particular employer in each individual case. But in the thought experiment we can offset the bias by reducing employers' yearly tax bills by A$10,000 for each employee, raising the general level of tax to compensate. Taxes only look higher, because after the rebates the tax bills stay the same - and they would drop as unemployment eased. Less tax wouldn't hurt the government, because the Social Security bill would narrow in step with the tax base. Employee rebates would encourage apprenticeships, discourage retrenchments and bring the long- term unemployed into the workplace by privatising Social Security.

The point about a tragedy of the commons is that the point of application (point d'appui, Schwerpunkt) matters. Negative Payroll Tax is just like Negative Income Tax (or the Australian concept, tax credits), except that it has to be applied to taxes paid directly by employers, i.e. production taxes - it wouldn't do to apply it to indirect production taxes, like property taxes paid by the ultimate owner. Incidentally, a separate thought experiment (which can be backed by mathematical modelling) shows that Negative Payroll Tax and Negative Income Tax converge asymptotically in the long run, but that is immaterial because here we are more concerned with the immediate response.

We are more concerned with the immediate response for two reasons. One is that in the long run we are all dead - a very slow acting "cure" is meaningless. The other is subtler. We have been dealing with a static situation, and the differential equations describing dynamic responses add complications all their own. Other external factors ("exogenous", in economists' jargon) come along - and if these isolated shocks come along on a time scale more frequent than three times less often than the main system responds, it never really gets close to equilibrium; the dust never settles. That "three times" figure is empirical, but it is backed by back-propagation in neural networks, by Shannon's work in information theory and by Heaviside's insights into how impeding telegraph signals actually improves signalling. (That last points up a flaw in making economic reforms to improve efficiency - sometimes the "inefficiency" is a necessary part of the drag to prevent wild oscillations.) If you do want a technical reference, look up the convergence of perturbation methods in solving rigidly coupled systems of simultaneous differential equations.

With a Negative Income Tax or similar we rely on an equilibrium process working much faster than economic shocks to restore an employment equilibrium. With a Negative Payroll Tax we get that automatically because it has a built-in instantaneous response - in fact there is a mathematical identity operating to keep Social Security costs in lock step with changes in the tax base. Technically this creates an invariant, a sort of automatic minimum wage, but the theoretical importance is that the differential equations can now be recast without one interaction that gives rise to instabilities. If introducing it overnight were too fast for political reasons it could be phased in - most conveniently with a grandfathering approach affecting cohorts born after a cut off date, to target youth unemployment.

Unfortunately the fact that we decide the invariant shows how this particular thought experiment is incomplete. We picked Australia as an example, and used the A$10,000 cost of Australian Social Security. Why not pick twice that, since that's closer to the Australian minimum wage? In fact, why not pick A$100,000? Surely there's a free lunch here.

Only, there isn't. We could try to pick an arbitrary minimum wage level, but that one wasn't picked arbitrarily. The level we picked doesn't just maintain the minimum wage invariant, it maintains a Social Security invariant too, keeping the system budget-neutral for other purposes. But even Social Security's not an objective level - it's picked for political reasons, which we can see because it's different in different countries.

However the underlying reason for Social Security is not wholly arbitrary and politically determined; it's the need to head off major unrest - the reason Bismarck instituted it in 19th century Germany in the first place, or why in Britain its predecessors the Poor Laws were linked to the Vagrancy Acts. Notoriously poverty is one of the causes of crime, so reducing Social Security below a certain point will eventually transfer the burden onto the budgets of policing and prisons. And there are resource constraints, so there's an upper limit - that's why we can't have a level of A$100,000. In the long run the limits can't be kept apart without sustainable demographic policies for Malthusian reasons. We can only pick any arbitrary level between the two limits. Fiscal drag automatically cures mildly high levels, but democracies could no more resist the temptation to set the levels too high than they can with agricultural support prices.

Now, we have found evidence that points at this tragedy of the commons mechanism linking Social Security, tax and unemployment. An honest enquiry has to look for counter evidence, and there is some - especially in the U.S.A. and Britain - but on closer inspection it is consistent with it, although clearly it justifies a further enquiry.

That evidence is that those countries have lower unemployment problems without having anything that works like a Negative Payroll Tax, and so did other countries before. Yes, and syphilis goes away if you ignore it - for a while. Disraeli's "Sybil" describes similar developments more than a century and a half ago, as did Marx, and the problems went away then too. The Negative Payroll Tax approach counters the effect of a reverse slope on a graph by applying an offsetting bias, like grid bias on a thermionic valve. That doesn't make it the only theoretical approach.

The simplest is to move along a curve by reforming or stepping up the economy - when it's booming the marginal costs change so more people are hireable even at given levels of productivity, even with the drag of the tragedy of the commons. That's what was implied by an earlier round of globalisation, the repeal of the Corn Laws in Britain just after the time "Sybil" was describing. Unfortunately this isn't sustainable if external factors keep moving the economy onto further and further contours with more and more of a reverse slope.

There is also the possibility that a NAIRU-like effect means that boosting the economy tends to shift productivity levels higher so you end up on the worse contours as a side-effect of staying out of immediate trouble. This is actually the logic of congestive heart failure, where lung congestion makes the heart work harder, which makes the lungs thicken from the increased pressure, which makes... until you die or something breaks the loop. This effect, if it exists at all, is related to the interaction of capacities and flows (integrals and derivatives, mathematically), and most economic material abstracts out and discards capacities as nearly immaterial. That means it would be hard to detect with standard methodologies, just as those discount the long-term stabilising effect of the Real Balance Effects, which do not express themselves because of all the other shorter term behaviours like the effects of intervention. Those stabilising effects aren't visible in the Keynesian model, for example, as that just works with flows. That may be one reason why all levels of economic activity fit that model unless you provide external, i.e. government, intervention - which only encourages the intervention which rendered any natural stability impossible in the first place, and may cause Pilot Induced Oscillation too.

The next simplest solution to the tragedy of the commons in unemployment is to put the clock back with a Luddite policy - that corresponds to moving from one contour to another with less of a reverse slope on it, so it is easier to be on the main slope. That doesn't work in the modern world, because you still have to compete with everybody else - even if you try to change the rules like the U.S.S.R. The contours with more reverse slope, of course, are those where the productivity of capital is greater and workers can no longer command the same real wages for their own marginal productivity - to that extent Marx was on to something, but it wasn't the whole story.

The last [approach to avoiding the problems of a reverse slope] is to privatise everything, even labour, so costs no longer get transferred away. That sounds good, but in plain language it means slavery, just as Negative Payroll Tax implies government control over who gets the work permits that bring the jobs out because the bureaucracy necessarily administers who qualifies for a rebate - strictly speaking, who qualifies to trigger an employer's tax reduction. There's a whole raft of relevant research material I have come across here, comparing the economic and social interactions of slavery in different times and cultures - but that's a whole other article.

Curiously, that [implied government control] doesn't apply to Negative Income Tax and the equivalents - we have a real world example in the Alaska Permanent Fund. That uses government funds but it is rather more hands off and subject to court review. If I were recommending a practical policy - say for Australia, which has a federal system - I would apply a Negative Payroll Tax to State level production taxes, or even make it a requirement of commercial leases of plant and property, an encumbrance compounding for tax liabilities structured to encourage such leases. Either way the objective is to distance the offsetting bias from bureaucratic or centralised control so the government could only regulate the framework, not [regulate the] individuals.

The discerning reader may note resemblances to some good ideas that didn't work, like Social Credit or Single Tax. True. Social Credit would work, if the initial deficit were used to build up a Perpetual Fund like the Alaskan one, with the first dividends going to wind back the deficit with a Sinking Fund approach - and if there were enough productive investment opportunities for everyone and the system were responsive enough. Single Tax would work if it tended to drive people and resources together, so that even if they weren't owned like slaves at least they were all owners. That could either work out like making us all rentiers like under the Social Credit approach, without the centralisation but with the unresponsiveness, or it could make us all peasants by giving us all opportunities to live like that at the same time as taking away all the opportunities to live any other way.

So there are some things we could do about unemployment, but there's still no such thing as a free lunch - just the lunch we already paid for and didn't get yet. No wonder they call economics the dismal science. I do see one big plus: we can boost the economy without inflation if we combine high taxes with high spending, especially military spending which trickles down better than most, because spending and taxing aren't exact opposites; up until now we couldn't do it for political reasons, but production taxes with a Negative Payroll Tax component are progressive and constructive without any direct personal impact except job creation, so they have a low or even negative political cost.

Return to the contents list.

Article printed in Defender, the national journal of the Australia Defence Association, of Spring 2002 (Northern Hemisphere Autumn)

- not precisely on economics

Connections between GDP and Defence

It is obvious that Australia's defence has been sadly neglected over recent decades. Very often that has been brought out by looking at the percentage of GDP going into it, either in relation to other government spending or in comparison with the spending in other countries.

That is a natural way to approach things if one is in the habit of looking at government activity and is only coming to defence issues with that as a toolkit. But it is of very little use in estimating just what sort of defence Australia is actually getting - how much bang for the buck. For that very reason, some people even deny that falling spending matters and make unquantifiable claims about how technological improvements are "really" giving a very different defence result.

So this essay will explore GDP percentage spending over history, and try to use those very different historical circumstances to bring out just what other features do or don't affect the defence results. In case anyone suggests that the comparisons are futile because (say) the world is very different since Napoleonic times, that is the whole point. By analysing what has remained the same and what has changed we will get a better perspective. And one thing the following quotations will bring out is how things looked to the writers in their own context, never mind how second hand their actual history may be; for instance I shall be quoting Gibbon to see how things looked in the 18th century, not how the Roman Empire was.

Here's Gibbon, near the beginning of chapter 5 of "The Decline and Fall of the Roman Empire", writing about the late 18th century: "It has been calculated by the ablest politicians, that no state, without being soon exhausted, can maintain above the hundredth part of its members in arms and idleness." - and he goes on to examine how that can be disproportionately effective when organised as suitably large and properly trained units.

There are two things to notice here: that 1% figure is in terms of tied up manpower, not a money measure of GDP, and it is assessing a sustained use, a regular standing army. Both are typical of that age, just before the Napoleonic era when there had been few technological improvements of a sort that needed capital rather than technique and when manpower correlated well with economic activity in the comparatively undeveloped economies of the day.

That is why there is no real contradiction with this quotation about the English Civil Wars of 1641 - 1651 (ish), from Buchan's 1934 life of Cromwell (near the beginning of Book II, Chapter I): "Even in the later stages of the war the total number of soldiers in the field was scarcely one-fortieth of the population." This shows us two things. To Buchan in 1934, this 2.5% figure - which was actually high - looked low. And it seems to contradict Gibbon's 1% maximum. The reason it does not is that even over a ten year period there were interruptions and that that is not a standing army figure; it falls nearer another ancient model of warfare, the tumultuary one. That is what the Romans meant when they said "tumult among the Gauls", apud Gallos - not that the natives were restless but that the clans were gathering. The last instance of this, with the fiery cross going through the land, was in the early 19th century when the Scots of Glengarry were called out and fought the southern invaders to a halt long enough for the regulars to come up, so that by their aggression the southerners gained not one inch of Canada.

That was no mere aside. It both shows us another model with much higher unsustainable surge proportions of resources committed, and it shows that money numbers aren't everything. Clan risings typically left one in five or six behind to mind the farms while the rest turned out, but the money cost was low - and the quality of the troops related to the link between their culture, their habits and expectations, and their tactics; read, training. Here is Nassau Senior describing the economic discrepancy in part of his early 19th century work on wages: "In the early stages of society, the rank and even the safety of the landed proprietor is principally determined by the number of his dependents. The best mode of increasing that number is to allow the land, which he does not occupy as his own demesne, to be subdivided into small tenements, each cultivated by one family, and just sufficient for their support. Such tenants can of course pay little rent, but they are enabled by their abundant leisure, and forced by their absolute dependence, to swell the retinue, and aid the political influence, of their landlord in peace, and to follow his banner in public and private war. Cameron of Lochiel, whose rental did not exceed 500 l. [pounds] a year, carried with him into the rebellion of 1745, eight hundred men raised from his own tenantry. But in the progress of civilization, as wealth becomes the principal means of distinction and influence, landowners prefer rent to dependents." Or as Buchan put it in Book I, Chapter I of his life of Cromwell: "The sixteenth century saw the breakdown of all the old relationships... landowners regarded their estates not as a nursery of men-at-arms but as a source of financial profit." - a change that worked through earlier in England, of course.

Well, just what changes did happen in warfare after the 18th century? It is not a coincidence or an academic exercise to ask about just that point of time, because that was the end of an era and the beginning of new ones with common features that last to this day - and may be ending with the Revolution in Military Affairs. Napoleon came on a scene in which tactics and weaponry had changed little in nearly two centuries. Things had come up against their constraints, and his genius was able to apply new techniques on top of these - but he did not actually displace them. The arms of war were the same, armies still marched on their stomachs, and so on - he just found new ways of working within those constraints, found them among the work of earlier theorists.

One change was to release the strict control of supply from organised lines - but not totally. (Interested readers may consult the 1911 edition of the Encyclopaedia Britannica about military supply and transport [continued here], valuable for a pre-1914 insight, which is available on the internet though I was able to consult a physical copy.) Another was to change the basis of conscription and bring in mass armies raised by mass conscription. Earlier, conscription had been needed in order to get manpower for long periods of service so the larger proportion of service could be after proper training; with discipline it made sense to get recruits from outside the homeland, since patriotism was unnecessary and that did not diminish the tax base. But Napoleon made "war feed on war", drawing down other areas' resources, and his mass armies had less training and so needed more motivation; he drew on France itself for men. That means that, even though Gibbon's 1% still applied, it wasn't France's own economy pro rata to that 1% except to the extent that French manpower was involved - and, with brief campaigns, in theory each class of recruits could be released in short order (that didn't work in his last days of rule).

Mass conscription brings problems of its own. It was imposed by mass billeting of troops, something that was complained of by the rebellious Americans; under the name of dragooning it was a well known technique. Draft evaders had nowhere to hide if their home village was occupied. But there was an economic cost. It is not often appreciated that French agricultural subsidies, which on the face of it appear pointlessly inefficient, were there to promote a peasant lifestyle; their output wasn't measured in profit or foodstuffs but in conscripts. It meant a shift in an entire way of life, one which is producing massive disruption now conscription there is ending (just as it did when it ended in Britain in the early 1960s). What this means is that it is hard to capture GDP numbers, and it is important to provide both financial and manpower numbers to get a true picture - and sometimes we have to estimate one from the other.

By the 1860s things had changed. Most Americans think "of course" the north was going to win their civil war, from its combination of superior manpower and industrial strength. But there was no of course about it; the world had only just changed, and indeed if there had not been compromises delaying that war there was a serious chance the south would have won. The thing is, the northern manpower was not that superior and actually represented a coalition of north and west (now midwest), which might have been split by more frequent defeats and so dealt with in detail. But industry was making a difference by then, particularly through mechanised transport. (Another change of that era was an increase in groups adopting guerrilla tactics; previously there had been insufficient strength available for their enemies to have garrisons in place to prevent insurrections going straight to armies in being, albeit untrained ones.) And, of course, more advanced economies and more sophisticated techniques of drawing on their resources - taxes etc. - meant that Gibbon's 1% limit on manpower and its implied limit on the underlying economy were increasing.

One thing that had not changed was the level of training needed for a rounded soldier, some two years of salting with real casualties in training; while it is possible to give more limited training for special aspects, such as skipping dealing with defeat or the way 1918 British soldiers were only trained for the trenches and not for the German breakthrough, all round training has not yet moved on. That is what today's problem is with rapid recruitment in the face of emergency, or with putting reservists straight into the role of regulars. This is something I shall return to.

Most of today's Americans think that the USA is militarily strong because it is backed by a strong military-industrial complex, and that this will simply continue. The American military know that this is partly true but that it is not the whole story. They are aware of training needs that include time as well as money, of how the economic connection is recent - and that it might change with the Revolution in Military Affairs, and that that revolution, while coming, has not yet quite come. Let's look at the connection as it was a quarter of a century ago, using 1975 figures in the "Book of Lists": in 1st place Israel had 5.53% of its population in the armed services, while 7th and 8th were North Korea and Portugal with 2.85% and 2.55%. Going back that far lets us use a little hindsight in seeing what figures like that mean - it lets us train our sight for looking at today.

The first thing is, there's a multiplier. There's a lot more GDP going in than that.

The second is, countries like Israel and North Korea aren't getting their financial costs covered simply by their own economies - they were drawing resources from their respective cold war sponsors.

The third is that it takes a while for things to work through; Americans confidently expected a rapid collapse of North Korea after the fall of the USSR, but it didn't happen that way. There were two reasons, the fact that they didn't have anywhere else to go - defeat was disaster - and that these things do take time to work through. We can use the Portuguese case to see that; it was fighting at least three colonial wars at the time, drawing on the home country's manpower when once colonies supplied manpower to it (in the 1960s the Goa garrison was largely recruited in Mozambique). While Portugal had fought Angola to a standstill so it was no longer a drain (and didn't have any paramount guerrilla group in waiting), and Mozambique was the largest single drain, there was an anomaly in Guinea-Bissau that was the real breaker. The Portuguese used a double-dip conscription, that brought back middle aged middle class professionals like doctors and engineers to support the counter-insurgency effort, and it was more serious in the smallest colony, Guinea-Bissau. It was represented that the way the Portuguese put up with it showed an underlying level of popular support for the regime - but in fact it did not, and built up a pressure in support of overthrowing it. Again, there was a time lag. The comparison with France shows that there is no problem with conscription if only most people have been through it and are not facing it - France has objections to ending it from people who have just served! But the main lesson here is that conscription was needed to get servicemen of the right calibre, where conscription in many other countries - in peacetime - was not so much to provide a standing army (as it had in the 18th century, though it did still do that) as to provide a pool of adequately trained reservists.

To the here and now. There is a change across the board, which we have to brace ourselves for. Once the Revolution in Military Affairs really does arrive, it will no longer need such a long support tail to work and it won't need a huge economy behind that. Does that mean that present low spending levels won't matter? Not at all, though it does mean it would be futile to invest in huge collections of obsolescent materiel now to make possible using fewer people later. And that has been our emphasis so far - to use capital and technology to stretch out other resources, particularly manpower. Capital and technology will correlate less well with each other, just as they have already ceased to correlate well with manpower - but there is still a connection. Spending levels would reflect comparative levels of defence, as between this country and others, as before, but we would lose any accumulated lead and have to rebuild it (something like this happened to British naval superiority a century ago). But it does mean that countries like Israel - which even as far back as the 1980s showed an American General in the Lebanon that it could have killed him with a drone - will no longer have an edge; inside a decade almost any country in the area, or even some groups, will be able to do that. The thing about mere technological advantages, as against capital ones, is that they can easily be superseded.

So Australian spending levels do have to be kept up, only the important issue is not just how much they should be but on what they should be spent. While spending forms a constraint on what we can achieve, spending alone does not achieve things - it can go astray. A simple increase in defence spending across the board, while it might revive quite a few things that are on the point of expiring, would as it were only go to fat and not muscle if it wasn't accompanied by the right exercise regime. So those are the things we have to decide - what the right exercises are. My personal gut feeling - totally without anything to back it up - is that 10% financial increases across the board would be useful without risking waste, while even 30% increases at regimental level would be useful that way. More would require detailed planning on how to use them before we made them.

Clearly manpower matters; in this day and age it is impractical to recruit outsiders, so it is important to have that large pool of trained reservists, ready to be upgraded to regulars at short notice even though they would always need some further training. Even if the training isn't produced by conscription, it will still cost money as well as time in itself and time taken from people's productive careers. But also the spending has to go on R & D, to get a locally supported technology base - you really can't buy it in, since the higher skill levels it itself needs preclude that working any quicker and cheaper than doing it yourself anyway (I've seen it in the computer area). Also, like stretching out manpower, it is possible to use capital in low technology ways to stretch out the available high technology - like using armoured cars in proper combination with tanks, though this again increases the load on skills and training since the resources have to be combined effectively in the field, which needs good junior officers and NCOs.

Taken all together, this shows some places we should look for our shortfalls and some things we would like to achieve to redress them. Now, how do we achieve them?

Return to the contents list.

Article not yet published - not on economics

Subroutine threaded code with back interpreted embedded data - a generic approach to implementing bytecode interpreters

Java, the web-oriented language from Sun, has been much in the news lately. One of its distinctive features is that it compiles into a portable bytecode, so its programs can run on any platform they are downloaded to. Any platform, that is, which has an interpreter that can run the programs. But Java is not unique in using this approach; BCPL - a predecessor of C - compiled to Ocode or Intcode, and of course UCSD Pascal had its p-system. In fact, whether a language actually requires it or not, an implementer may decide to use an intermediate code. This article explores some aspects of implementing these bytecode interpreters, whether for Java or for anything else.

Many compilers work by converting the source code to an intermediate form, then using that. Sometimes they convert the intermediate form to object code by using it to drive a code generator, in which case you may not even be aware that compilation went through the intermediate stage. Sometimes the intermediate form is interpreted, in which case it may still be hidden from you - the interpreter may be integrated with the operating system so it does not need to be run as a free-standing utility, like the batch file interpreter under MS-DOS. Alternatively, it may be hidden as a packaging issue: you can get what looks and behaves like an executable file, but is really a copy of crucial parts of the interpreter with the intermediate code to be interpreted attached as embedded data. Occasionally, however, there are practical reasons for not packaging them together like this - for example, if you compile to an intermediate form and stop there, you can port it to the platform of your choice. Java is one of these cases. This makes the intermediate form very obvious, and it is usually best to interpret it as so many of the behaviours are clearer at run time than at any earlier stage. Since these forms need to be handled on a wide range of platforms, there is little point using a highly specialised format for them as that would only make the run time work more awkward on those platforms that it didn't suit; and the designers of these formats have often settled on "bytecodes", that is, intermediate forms where the intermediate instructions are organised around bytes as units.

This is where we come in. Before we can port a bytecode program to a new platform, we need an interpreter available for that platform, so there is a need for a generic approach to implementing one. If we want to implement an interpreter for the bytecode form generated by some compiler we want to do the best job possible in terms of simplicity and robustness, and as a second priority we want it to be efficient. If efficiency were our main priority we would probably go for a code generator, but that would give us a huge load in terms of testing it and supporting it and we could not port it readily from one platform to another - and ease of porting is one of the main reasons for using bytecodes in the first place. But how far can we take a generic approach to designing an interpreter? That is, just how much can we do to increase the efficiency of memory use and execution speed of an interpreter before we commit ourselves to some details of the host computer's own architecture? Quite a lot, as it happens.

The first thing we do is, we cheat. The obvious way to design a bytecode interpreter is to code it roughly according to the following pseudocode:-

Initialise virtual machine
Load current bytecode
Action it (maintaining instruction pointer)
Closedown actions.

This works, but there is the overhead of the loop and a case statement to action the bytecodes (although we could use a lookup table to tune the case statement). If we happen to be implementing virtual memory ourselves there is the overhead of maintaining that, keeping the right pages in memory on every cycle so as to keep the next instruction available. We can cheat because nearly every computer architecture supports subroutine call instructions of a fixed size - this is important - and even those that don't can emulate them with a very few inline bytes (we can use branch instructions for the calls, making sure each subroutine ends with support code - an "inner interpreter" - to maintain a return address to transfer control back to). The way we cheat is we pass all the work of interpreting to the hardware and we eliminate the inner loop completely. We make our interpreter load routines convert the bytecodes as they load them, treating them as macros for call instructions to action them; this can be done efficiently at load time. This does increase the demand on memory, but that is not usually a difficulty these days, and even if it is there is a convenient way to implement virtual memory too. After loading, running is simply a matter of branching to the start point. Branches and such need to be able to access the addresses of their destinations, which are held as embedded data; action support subroutines for these need to be able to access return addresses in order to reach the embedded data, which is not usually a problem. (If it is, we again have to implement our own subroutine calling and returning mechanisms.) What we have got here is subroutine threaded code. We have actually put together a dumb code generator with the help of 256 action support subroutines - one for each bytecode. It's dumb because it hasn't optimised the code and because it has foolishly converted everything to code, even the data. The most important embedded data will be branch addresses, but there are other things in there like numbers and text strings. Have we painted ourselves into a corner?

It turns out that encoding the data doesn't matter. Unless we code synonym bytecodes with the same calls or treat some with no equivalent in our environment as no-ops there is a one-to-one correspondence between the call instructions and the original bytes, so we can always back interpret at run time to recover the original data. For context-independent data this is all we need, but things like branch addresses are context dependent; that is, skipping the next three bytes needs to be converted into skipping the next three call instructions and so on. Luckily the original bytecode worked in standard byte sizes - this is one of the reasons for using a simple intermediate format - and the call instructions are also of a standard size, so the context dependence only means a fixed scaling. There is no problem.

Hold on, maybe there's no problem in principle, but isn't there just as much work doing this back interpreting as there ever was in an ordinary interpreter, and doesn't that mean there's a problem in practice? Even if that were true, it would only apply to back interpreting perhaps half the call instructions, which is a great improvement over having to interpret most of the bytecodes, and there is still no inner loop. (Of course, if we have to emulate our subroutine calls that is equivalent to the overhead of an inner loop - but no worse.) But it turns out that there are several ways to back interpret efficiently. We may be lucky enough to have a computer someone designed to make this sort of thing easy; we just might have 256 spare interrupts and our call instructions would have a regular correspondence to the original bytecodes, so we could back interpret by just masking and shifting the call instructions. Even if we have to use ordinary call instructions, which is far more likely, we can still fake this regularity with a jump table: instead of putting the action support subroutines where they fall naturally, we make each begin with an unconditional branch of standard size, and we arrange these in a table; they simply transfer control to wherever we find it convenient to put the main bodies of the subroutines, which terminate with subroutine return code in the normal way.

That is a clean and simple way to back interpret embedded data, but it has the cost of making all code execute even more indirectly, adding the overhead of branching through the jump table to every action. There is a better way. We can make each action support subroutine relate directly to its own back interpretation by laying it out like this:-

BYTEnn: 0Xnn

With this layout, all back interpreting "call ACTNnn" needs is to mask out the opcode for "call" and shift the result if necessary giving the data ACTNnn, subtract 1 (the size of a byte) giving BYTEnn, then load the byte that was stored there. We can usually tune this further by insisting that each BYTEnn start on an even address (if necessary wasting some memory), and in the same segment as ACTNnn if there is a segmented memory architecture, so ACTNnn only differs from BYTEnn by having the low order bit set and the masking can handle the logical subtraction. We can still do something like this even if code has problems with needing to start on word boundaries (e.g., just make the last two bits of BYTEnn both zero, and the last two bits of ACTNnn one and zero), and we can generalise it to hold several different versions of the back interpreted byte if it makes sense to store scaled versions to help with context-dependent back interpreting.

One generalisation that is very useful is to make back interpreting object-oriented. That is, we use an object-oriented style of machine code to do the back-interpreting indirectly - we do not usually have an object-oriented assembler to help us do it. We use this layout:-

BYTEnn: 0Xnn

The subroutine at LOADnn is of standard size so ACTNnn is still at a standard relative position, and the code it implements is probably quite short - if necessary we can keep it short by having a branch to its main body. Effectively, back interpreting can now be done either directly and efficiently by looking up BYTEnn, or indirectly and flexibly by means of the method held at LOADnn. Typically, code to load the byte either directly or by calling the method relies on the value of the return address of the current action support subroutine, and advances it. This allows several consecutive bytes of data to be accessed and lets the action control subroutine terminate and pass control to the next one properly.

Getting back to the action support subroutines, if we have enough memory we probably want to write them using only inline code for efficiency. If there is a shortage of memory we can conveniently build them up using calls on utility subroutines to do the back interpreting. Similar calls could also implement any general functionality of the language we are supporting, e.g. garbage collection or task switching. In fact, this helps us decompose the work: if we want to break it down for a team, we first prototype this general functionality as utility subroutines; then we give small teams the work of implementing related groups of action support subroutines using these utilities; then we integrate them to give an alpha version, and before we go into beta we decide which if any utilities should be expanded out as inline code. The utility subroutines should avoid using local variables for efficiency and to make it easier to inline them later, of course. But even without a team effort, building up an interpreter is probably within the capabilities of a single programmer working over the course of several months.

Strictly speaking that is all we need for a generic approach to implementing bytecode interpreters, but there is more we can do: back compiling embedded data, and supporting virtual memory (it's usually too hard to do both).

Back compiling means that we don't have action support subroutines for, say, pushing a literal value to the stack, but instead we have ones that compile themselves on the fly. That is, instead of a sequence like this:-

call ACTNnn /* back interprets next call and pushes it to stack */
call ACTN0A /* data for decimal 10, wrongly converted to a call */

- we still have the sequence, but now the subroutine at ACTNnn does something else. It patches the calls - both the real one and the wrongly converted one - to native code with the same action, which of course includes an embedded 0X0A for the decimal 10, and pads the rest of the space with no-ops; then to carry out the action of the bytecode it does a return to the original call point (which now holds the compiled native code) rather than to the next code. This back compiling only ever happens the first time, as on subsequent passes the compiled code is already there.

One obvious problem is that the compiled code might not fit. Usually it will, as the original space had to hold call instructions with addresses and the data itself usually becomes more compact. There may however be a problem with squeezing no-ops into the leftover space, but even that can be worked around if we are willing to settle for only doing half the work. That is, back compiling generates this sequence:-

call ACTNxy /* pushes next byte to stack */
0X0A /* data for decimal 10 */

- which still has an interpretive action, mainly to skip over the garbage, but does not need to back interpret the data it has to push to the stack. It would be really amazing if this took up more space than the wrongly converted version. Of course, ACTNxy is the address of an action support subroutine that does not correspond to one of those the bytecodes generate calls for.

Why can't we do this conversion at load time, and maybe skip the necessity for padding shorter code sequences with no-ops while also allowing longer code to be put in inline? It would even let us put things in ROM, which we can't do with back compiling. In fact, why not put in all action support code inline? We could even go further and optimise the joins between pieces of support code. Of course we can. The only price we would pay is making the loader smart enough to work out whether each byte is code or data, optimising the joins, adjusting addresses on the fly, and possibly code bloat. In fact, we would be committing ourselves to implementing a code generator. Leaving back compiling until run time guarantees that we know exactly what each original byte was for, and we can generate the correct native code for it, right?

Wrong. Just because we use a bytecode one way on one occasion, that doesn't mean that is the only way we want to use it. We have come up against one of the issues involved in choosing early or late binding. To give just one example, consider the methods in an object-oriented language: these are often implemented by giving each object a pointer to a table of methods for its class, and that table in turn holds pointers to the code implementing the actions of that class, or to entries in the tables for other classes if there is inheritance. The tables might really implement this as pointers, or they might be jump tables as discussed earlier - each table actually holds executable code with the responsibility of passing the buck until real activity takes place. This is a blending of handling the methods as data and as code - in fact, that is what object-oriented programming is all about. While we can handle this sort of thing, we need to take a great deal of care and be absolutely certain we have considered all the cases. It is usually not worth it, since our priority is simplicity and robustness; back interpreting is fairly efficient anyway. Luckily, we don't have to commit to back compiling everything - we can pick and choose what to go for, and indeed we could handle implementing back compiling as a series of upgrades as part of a development cycle. In particular, we may find that back compiling branch instructions into native code is worth it because that tunes loops, but we need to be careful.

There is another refinement to back interpreting, which is a lot more reliable to implement and certainly worth having. This is virtual memory. The only thing is, these days we may have it already or the need may have been bypassed by having plenty of memory - although there is no such thing as enough. But in the good old days when life was tough programmers often had to fend for themselves or do without, which is why many versions of Forth implemented their own virtual memory (and why many people resorted to Forth - it is a portable systems language you can implement yourself and that fits into computers with only small amounts of memory, and it is a vast improvement over nothing at all). Databus, the proprietary language on the old Datapoint, also implemented its own virtual memory. Both of those families of interpreters could provide multitasking, too. Who knows, you may be one of the people involved in providing these things for others. With all that, this may be of more than just academic interest.

The first thing we need for virtual memory is a set of utility code to handle reading and writing blocks of data between disk and buffers in memory. With any luck our operating system not only provides this but also manages the buffers on a least recently used basis. If we have a really rigid operating system we have to specify the buffer handling ourselves (or our compiler does), so it is worth giving an overview of one way to implement handling them on a least recently used basis. If we want, say, five buffers we set up a management table with five entries. Each entry has an item specifying a different buffer and another specifying which block it holds, and whether it needs to be written out to disk. This would usually be because it had been altered since last being read or written, but it might have been preloaded with the interpreter. Apart from that, initially each entry specifies that the buffer is unused, perhaps by holding a dummy block number. When a block is needed a backward linear search of the management table looks for it. If it does not find it already in a buffer, a forward search looks for an unaltered entry, reads the desired block into it and maintains the table, then uses the buffer. Failing that a forward search looks for an altered entry to reuse, only its buffer is written back first. Unused buffers are automatically picked up during searches for unaltered ones. Whatever happens, unless the very last table entry was found the entry is swapped with the next one so the least recently used entries migrate towards the front and are more likely to be reused first. If there is any idle time during the interpreter's function, this can be used to write back altered buffers and mark them unaltered. At the end of the session we can discard all the buffers unless we are implementing persistent data objects, in which case we have to write back all the altered entries.

Another reason we might want to implement virtual memory ourselves is that we may want to keep the disk version in bytecode form. We can convert bytecodes to calls on reading blocks in, and vice versa on writing them back out. That gives us a degree of compression for free because we have inside knowledge of how to do the compression.

For virtual memory itself, we do not need to change the code that is generated - all the call instructions - but we do need to change the load routines slightly to put the generated code into the blocks. We should choose block sizes so call instructions do not get split between blocks. We also need to change some of the action support subroutines, but that is a straightforward matter of writing their code to call utility subroutines or methods to maintain the management table and page blocks in and out, and also to convert virtual memory addresses into real addresses within buffers before accessing memory locations. We do not need to use any specialised interrupts or exception handling. Much of the need to access virtual memory matches the need to back interpret embedded data, which we probably want to implement as calls on utility subroutines to save real memory. The programming effort can share the overhead of that and there will only need to be a very few calls on the virtual memory support utilities, since most can be strategically placed within the back interpreting code.

Surprisingly, a great many action support subroutines do not need to be changed because they do not have to work with virtual memory, and that means that they do not need the overhead of paging buffers around - unlike a simple interpreter, which needs to do it on every cycle. For example, the simple C statement

x = 2+3;

might generate bytecodes like this pseudocode:-

Push next byte to stack /* needs to access embedded data */
Push next byte to stack /* needs to access embedded data */
Add top two stack entries
Push next word to stack /* needs to access embedded data */
&x /* address of variable x */
Assign /* needs to access virtual memory location */
Clear stack /* because a C assignment leaves a value */

The code for addition and clearing the stack does not need to access virtual memory, unless we want to implement that in virtual memory too.

What happens when control reaches the end of a buffer? Don't we need to check the virtual memory on every cycle anyway to prevent falling off the end? No, because we can put sentinel code at the beginning and end of each buffer to catch it - it's just like catch and throw exception handling, only we don't even need to support any exception handling; we just let control fall off the end. The sentinel code works out which virtual memory address should have been reached next, pages it in and branches to it. We would usually want the more flexible object-oriented approach to back interpreting. The sentinel code would also have a method stored with it; that would also roll in the corresponding next or previous virtual memory address, only it would inherit its method instead of actioning its action support subroutine. It would also have to patch the return address of the current action support subroutine to ensure proper control flow. We might have difficulty squeezing the sentinel call in between buffers, or the sentinel method in before its action code, but in each case we only need a single branch to the main body of it. If the operating system is too rigid about where it lets us put buffers we can still work around it: just put a special call on a virtual memory handler subroutine at the ends of each block each time we load one, so the sentinel is actually within the buffer. This does not count as part of the virtual memory address space, so it is invisible to the normal functionality of the interpreter. We can even do this if we only convert bytecodes when modules are initially loaded, with calls held uncompressed on disk. Doing this is wasteful but workable. Probably the most awkward thing that can happen at the end of a buffer - or rather at the end of the code before a sentinel call - is when a branch occurs. The call on an action support subroutine for the branch is at the end of one block, but the embedded data for its destination address is in another and the address itself might be in yet a third. If the calls on utility code have been coded properly this will work itself out, but we have to be careful to test this sort of thing and we ought to have enough buffers to prevent this giving thrashing. The complexities involved in testing all combinations are why we shouldn't try back compiling if we are providing virtual memory - after all, this kind of branch would be I/O bound anyway.

Finally, there are some conveniences - bells and whistles - we could provide. We do need to provide some form of user interface, or the interpreter will never know what to run. If the language we are interpreting is enough of a systems language we could provide a default startup module with the responsibility of working out which application to load - perhaps by interpreting a command line, perhaps by working out which icon was selected, perhaps by having a dialogue with the user. It could be preloaded, either giving a virtual memory work file already holding calls on action support subroutines or giving an executable file which is really interpreting its own data as described earlier - or a combination, if the startup module could fit in an executable file in the buffers provided. That would mean that we could have virtual memory without having to maintain a separate preloaded work file, so there would not be the problem of version control. The startup module could also work out the host machine configuration including any work files to use for virtual memory. It could even do the conversion work of loading new modules and converting them to calls on action support subroutines once and for all, if we don't use virtual memory. Since this amounts to a sort of stub operating system on top of the host operating system, we might be able to include the source code compiler that generated the bytecodes (if that was written in itself, as Oberon is) - although that would probably not fit in one executable file. We could even give the stub operating system the functionality to create executable files out of other bytecode programs, although those would be restricted to the one platform unless they were put through a recreator module to decompile back to bytecode - simple enough, since we could use the back interpreting to do it. In fact, Oberon can even support its own entire operating system, although we might not want to go that far.

Another thing we could do is take a leaf out of the book of techniques used elsewhere to help porting. For instance, eForth is an implementation of Forth that was put together in assembler in a way that makes heavy use of macros and boilerplate or template code rather than direct use of assembler. The idea was to have templates of machine code, either in assembler source code for the host machine or as bit patterns held as data giving machine code for other hardware; those could be changed according to the target machine. Most of the generated Forth was built up from combinations of these templates, or pointers to combinations of them; the templates changed from machine to machine, but the combinations didn't. The fact that the assembler normally generates code for its own host is irrelevant, except that it knows about allocating labels and addresses. If we use this approach to generate a bytecode interpreter, we could provide an interpreter for each subsequent new machine quite quickly. In eForth the idea was to use an assembler for the job, avoiding using Forth to host generating the new systems because of its learning curve and because that needs a host system which already has Forth. If we have a host environment where the interpreter is already implemented we could use our own language for it - Java or Oberon or whatever. We would need to use another approach for the very first target environment, either writing it the assembler way or in another systems language. The result would be the opposite of a compiler; instead of varying the source programs and targeting a fixed environment like a normal compiler, this code generator would only ever generate one program - the bytecode interpreter - but would be able to switch to new target environments quite easily. Datapoint did something like this for Databus, although that only targeted one machine; they did not release a standard interpreter, but an installer program. This was given a configuration specification, including how many multitasking partitions were needed for different data entry terminals, then generated an interpreter for that.

That provides a fairly complete agenda for implementing bytecode interpreters. Perhaps one day someone will provide a cheap commercial Java or Oberon implementation using it - that would be better than a public domain version, as only a commercial product can reach a wide user base. Public domain products can remove any profit margin and make commercial ones impractical.

[Afterword: on many platforms direct threaded code will be at least as efficient as subroutine threaded code.]

Return to the contents list.

Back to the home page.