Insurance problems

Christchurch is going to have to face higher insurance costs for a while. Earthquake risk here is now revealed to be higher than we had previously expected. We're now less likely to get major quakes than we were a year and a half ago, but we're probably more likely to get them than we had expected we were as of a year and a half ago. And so the risk payment goes up not because the actual risk has increased but because we have better expectations of the real risk.

John Pagani argues for the nationalisation of AMI, the big Canterbury insurer that had to be bailed out when it was revealed to have insufficient re-insurance for the series of major Christchurch earthquakes. And he makes some reasonable arguments: if the government's going to be liable for bad outcomes, it ought to have some hand in making sure it doesn't happen again. And, he's right that there are moral hazard problems with the big private insurers perhaps banking on the potential for bailout; I know that I didn't worry about checking into AMI's asset and risk structure because they had so substantial a Canterbury presence that, even if they tanked in an earthquake, there was no way the government would fail to bail them out.

But I'm not sure what problem nationalization solves. You can make a good case for that companies with exposure to concentrated correlated risks buy more reinsurance than those having a diversified customer base, and that regulations could be tighter around that. But whether AMI is private or public, it will still have to buy reinsurance on the global reinsurance market. And if that market has seized up, I'm not sure how changes in AMI's public or private status affects things.

I remain a bit perplexed about why people can't get new policies. You can't insure against a certain risk, but if the risk of another February hitting Christchurch is around 5%, you'd think reinsurers would be happy enough to issue cover at a fairly high premium. Some potential explanations:

• Reinsurers fear being saddled with the costs of prior quakes in any new event if a full assessment of a property's prior damage hasn't been completed
• But then, why is there difficulty in getting coverage for new builds?
• Uncertainty over what portion of future claims will be covered by EQC if the EQC fund is exhausted,
• But then, wouldn't we expect solution through better insurance contracts? 
• Reputational costs of actuarially fair pricing would swamp potential returns
• But reputation accrues mostly to the local agent, not the big reinsurer; for those, reputation is determined, I would have thought, by track record in paying out.

I think we need a fair bit better understanding of what's going on before we start nationalizing insurance companies. 

The case against a dirty GST: Illusory benefits of exclusions.

John Pagani, in the blog post that motivated my post on the GST yesterday, suggests four reasons for supporting the exculsion of fresh fruit and vegetables. In short, these are that it would mitigate the regressive nature of the GST, that it would promote better nutritional outcomes, that we already exclude a number of products from the GST so it is already dirty, and that the best thing about the policy is that everyone gets the same break. Let’s take each of these in turn.

1. Excluding fruit and veg is a way of making the tax less regressive.

Actually, John didn’t exactly make this claim, but it is implicit in his analysis, and it is a common mis-concpetion. First, please shout if from the rooftops: The GST is not a regressive tax. Steve Landsburg had a good post last year , which I won’t duplicate here, explaining how a flat-rate labour income tax with no tax on capital income (which is equivalent to a clean GST) implies a flat rate of tax. Adding in taxation of capital income results in double taxation. You can argue that a clean GST is undesirable because it doesn’t have the double taxation of capital income (compounded by inflation), and such double taxation is a good thing given that the rich earn a greater fraction of their income from interest and other capital income than the poor, but don’t say that a clean GST is regressive. Furthermore, if you want to argue that a clean GST is bad because it doesn’t imply double taxation of capital income, please don’t also advocate Kiwisaver or other tax incentives to encourage retirement saving. You can’t have it both ways.

More important, when one considers the overall tax system rather than just the GST in isolation, it becomes clear that excluding a class of goods from the GST is a lousy way to achieve income redistribution. To illustrate, consider an economy with a broad-based GST, but one that includes an exemption for a particular class of goods done for equity reasons rather than to encourage consumption of those goods. Now consider an alternative system in which the exemption is removed, and the extra revenue earned is returned as a negative poll tax to every adult in the country. Such a redistribution would be easy to implement. It simply means bringing the implementation of the benefits (dole, superannuation, DPB, etc.) into the same system as income tax, and then either increasing benefits or reducing the tax liability by a fixed amount. This policy will eliminate the distortions implied by having differing rates of tax, and would eliminate the compliance and enforcement costs that come from having different tax rates across different goods. And the impact on equity? As long as the rich spend a greater absolute amount on the good that had been zero rated than the poor, even if the good is a necessity that takes up a smaller fraction of their income, this policy will see the tax bill of the rich being raised by a greater amount than the poor while all receive the same rebate back. That is, removing a GST zero-rating would be a distortion-free way of taking from the rich to give to the poor! Only if the good in question were one where poor people spent a greater absolute amount than the rich would removing the zero-rating not be equity enhancing. Are there such goods? I haven’t seen data on this, but at a guess the only candidate goods would be fast food and cigarettes.

2. Exempting healthy foods is desirable/necessary to achieve good nutritional outcomes.

There are two issues here: The first is whether one wants to use Pigouvian taxes and subsidies to bring about behavioural changes in the nation’s eating habits; the second is whether GST exemptions would be the best way to do so. Let’s assume that we do want to use tax interventions. In this case, the optimal policy would be excise taxes on unhealthy foods and excise subsidies on healthy food. The nutritional value of food doesn’t change just because it is used as an intermediate good; apple slices don’t cease to be healthy just because they are served at McDonalds as an alternative to fires in a happy meal. (Note to American readers: the apple slices in New Zealand happy meals are not peeled nor served with caramel dipping sauce!)

3. But we already exempt goods such as existing houses and financial services.

This betrays a misunderstanding of the difference between exemption and zero-rating. Financial services are exempt, not zero rated. This is done to avoid definitional problems, and doesn’t cause the increase in compliance costs that zero rating does.

For second-hand goods, recall the principle that a GST is designed to look like an income tax without the double taxation of those who delay their consumption. Consider a world with a 50% income tax and no taxation of capital income, and then replace it with a 100% GST on newly produced goods and services but not on second-hand goods. It is a trivial exercise in general-equilibrium theory to show that these two worlds are identical. (Exercise for any undergraduate economics students reading this—show that the after-tax prices of both newly produced and second-hand goods will double in this example, even though second-hand good trades are not subject to the GST.) John’s statement that “a new house costs 15 per cent more than the identical house across the road that’s a year old” is simply wrong.

The bottom line is that none of these examples of goods being exempt in any way implies that we have already stepped on to the slippery slope of a dirty GST. Zero-rating fresh fruit and veg would be such a step.

4. “The best thing about the policy is that everyone gets the same break”.

You have to be kidding me. People like me who spend a lot on fresh fruit and vegetables would get a big break; those who meet their nutritional needs from canned and frozen fruit and veg would not. If you want to advocate for a tax break for the middle class at the expense of the poor and the rich, there are more honest ways of doing so than supporting exclusions from the GST that would cater to the consumption patterns of the middle class.

The case against a dirty GST: Some GST basics

This blog post by John Pagani about removing the GST from fresh fruit and vegetables has raised an interesting issue about value-added taxes like the GST and how well understood they are. Pagani comments; “I doubt there is anyone who would support a GST of 20 per cent with no exceptions; that would be inhumane”. Well, I don’t think of myself as inhumane, but if we had a 20% GST I would still be in favour of it applying as broadly as it does now. In fact, I would make the opposite claim to John: No-one, no matter what their political perspective, their desire for income redistribution, their concern about unhealthy eating, their belief in the efficacy of price signals as a way of changing eating habits, or even their basic level of humanity should every be in favour of muddying the GST by removing it from particular categories of goods. The key word in this sentence is “should”. I am convinced that the reason there are differences in opinion on this issue is misunderstanding about the GST.

In this post, I want to point out some aspects of a GST that are not as well understood as they should be, and in a subsequent post I will explain why commonly stated justifications for giving special treatment to some goods just don’t hold water.

The first thing to note is that a clean GST is equivalent to a labour income tax at a flat rate with no tax on capital income. By example, imagine an economy with a 20% flat-rate income tax on labour income and no tax on capital income. Now replace it with a 25% clean GST. This implies that 20% of the cost a product is the GST, hence the equivalence to a 20% income tax. In the income tax system, consumers who save have 20% of their income taken off at the point they earn it, and their savings then grow at the market rate of interest. With the GST, nothing is taken off at the start, the full value of their income grows at the market rate of interest, but then 20% of the total value is paid as GST when the income is finally consumed. Either way, consumption is reduced by 20% by tax. In short, the two systems are identical in the impact on consumer’s consumption possibilities and in government revenue.

Second, it is a basic mistake in tax policy to examine the efficiency and equity aspects of a single tax in isolation, rather than considering the effect of the overall tax system. For instance, the distortionary effects of a tax on a particular class of goods depends on whether there is also a tax placed on the goods that people might substitute too. Similarly, the equity implications of a GST depend on the income tax system that exists at the same time. Consider the tax system we had for most of the period of the 1999-2008 Labour government: a 12.5% GST, and income tax rates of 19.5%, 33%, and 39%. Now consider a change to a 20% GST. To maintain the same level of progressivity (and humanity) in the overall system, all that is needed is to increase benefits by 6.6%, and reduce the respective income tax rates to 14%, 28.5%, and 35%.

Finally, it is important to distinguish in a value-added tax like the GST between exemption and zero rating. The idea of a value-added tax is to tax final consumption but not the use intermediate goods. (If all sales were taxed, we would create a tax incentive for massive vertical integration, probably at a cost of bureaucratic inefficiencies.)

One can try to tax final consumption by guessing which purchases reflect final consumption (such as a retail sales tax), but this will not catch all final expenditure and will catch some intermediate-good purchases. A value-added tax works by having all sellers add the tax to the price of their goods but having them subtract off the tax already paid on intermediate goods. In this way, it is only the seller’s value-added that is taxed at each point in the production chain, but consumers will face the full value of the tax wherever they purchase.

Let’s work with a simple example with a 10% GST. A miller produces $10 bags of flour, to which it adds $1 GST, giving a cost of $11 to bakers and final consumers. A baker combines that $10 of flour with $10 of labour to produce $20 of bread, to which it adds $2 of GST. The baker, however, only pays $1 to Inland Revenue as it subtracts off the $1 already included in the price of the flour and paid by the miller. Finally, a café combines the $20 of bread with another $10 of labour to produce $30 of sandwiches to which $3 is added. The café subtracts off the $2 tax already paid by the miller and baker, and sends the final $1 to the government. The price to consumers of any of these three goods ($11,$22, or $33) includes a full 10% of tax.

Zero rating is typically applied to a category of good rather than to a category of seller. Using the same example, let’s imagine that bread is zero-rated, but flour is not, and nor are café sandwiches. The miller seller sells flour to the baker for $10 + $1 GST. The baker combines the flour with $10 of labour to produce bread worth $20. It places no GST on this, and sells it for $20 but claims the $1 paid by the miller back for a tax bill of -$1. The café still adds $3 to the price of sandwiches, but as the bread had no tax applies, it can deduct tax already paid and so remits the full $3 to the government. Consumers therefore pay the full 10% tax if they buy sandwiches or flour, but not if they buy bread.

Exemption works differently, and in New Zealand it is typically applied to the seller rather than the good. Exemption means that you don’t have to add GST to the price, but you also can’t claim back the GST already paid. Exemption is done to reduce compliance or implementation costs rather than to materially affect the price of a particular good. For instance, small businesses with turnover of less than $100,000 can choose to be GST exempt, although for some there is an advantage to registering for GST anyway. In our example, if bread were exempt, but not flour or sandwiches, the miller would charge $1 for the flour. The baker who added $10 of labour would then have to sell the bread for $21, so that bread-buying consumers would still be paying an effective 5% tax. And the café adding $10 of labour to $21 bread to create sandwiches would have to charge 33.10 for the sandwich, being unable to claim back the implicit $1 of tax paid by the miller, so that the implicit tax rate on sandwiches would be 10.33%.

This difference between zero-rating of goods and exemption of sellers, is the key to understanding why New Zealand’s clean GST has far lower compliance costs than the typical value-added tax. When completing one’s GST return on-line, one can simply enter two numbers: Total sales to domestic buyers, and total purchases from GST registered sellers. Because there is a single rate of GST, that is sufficient to calculate the GST component of your sales and the GST already paid on purchases. When some goods are zero rated, the fraction of the price paid on goods that is GST will depend on the proportion of each purchase that was for a GST-rated good. Filling out returns then means adding up the GST component of every single purchase separately. It is true as Pagani says that “the purity of a flat rate is good, but it’s not the whole enchilada.” Maybe this increase in compliance costs would be tolerable if abandoning purity brought other benefits. But the putative benefits of exclusions are illusory, based on mis-understanding the above points.

More on that tomorrow.

I'd like to thank the Academy...

Really, I should be thanking Luis. Here's the University of Auckland Stats Department's inaugural winner for bad stat of the week:

The nominations were all fascinating for a variety of reasons and much could be written about each of them. We’ve chosen Eric Crampton’s nomination of John Pagani’s heated blog post on youth unemployment:

In the midst of extensive discussion of the rise in youth unemployment starting around Q4 2008, Pagani points to changes in apprenticeship funding as a policy shift that could have generated the change (arguing against changes in the youth minimum wage as having been the cause). He writes:

“If it wasn’t the removal of the youth minimum wage that caused youth unemployment to increase, then it would have to have been caused by something else that happened around the same time.
One other big change was the a sharp fall in young people getting skills for work.
In December 2008 there were 133,300 people in industry training. By the end of last year, there were 108,000. ”

You could be forgiven for assuming that about 25,000 kids had been kicked out of apprenticeships – it sure looks like he’s referring to youths. All the other discussion is on youth unemployment. But the number he’s citing is overall enrolment in training and apprenticeships. And the drop in youth enrolment in training – about 4,000 – is nowhere near large enough to provide a plausible alternative explanation.

Congratulations Eric!

I'm going to point here next time somebody puts up changes in industry training as having caused the big increase in youth unemployment.

I'm looking forward to nominating the social costs of alcohol or tobacco when next somebody cites the number in press....

Mendacious use of stats [updated]

John Pagani suggested changes in apprenticeship funding could have been responsible for the increase in youth unemployment that I've ascribed to the abolition of the differential lower youth minimum wage. It seemed a plausible explanation. But one of my helpful commenters, Luis, links through to the actual stats on apprenticeship uptake.

Pagani cited numbers on total enrolment in apprenticeship programmes. Those figures seemed really high if they were meant to represent 15-19 year olds. There's no way that a third of all people in that age bracket were in apprenticeship and industry training programmes. TEC stats, courtesy of Luis, show 13,859 15-19 year olds in training as of 31 December 2008 and 9,657 in training as of 31 December 2010. So roughly 4000 fewer kids are in training programmes than were in training at the end of the last Labour government. But excess youth unemployment is now around 13,000. And, I'd be shocked if apprenticeship figures weren't fairly pro-cyclical - the construction industry, among others, will be less willing to take on apprentices when the economy's doing poorly. At least some of that 4k drop would then come down to economic conditions that have already been accounted for in my simple regressions.

If we had TEC data going back farther than December 2008, I could run some simple projections of what we would have expected training numbers to be given the recession. But it doesn't, so I can't.

Here's what Pagani said:

One other big change was the a sharp fall in young people getting skills for work. In December 2008 there were 133,300 people in industry training. By the end of last year, there were 108,000. No wonder unemployment has gone up.

So he doesn't explicitly lie by saying that the 133k represented youths in training. But he sure makes it easy for lay readers who don't know there are only about 320k people total in the 15-19 age cohort to assume as much. And, despite it being a blog post rather than a print newspaper column, he doesn't link back to the TEC stats that are the source of his figure so that folks can't easily check things for themselves. Pretty smelly, John.

Update: And, of course, there's the obvious point that if the changes hit both adults and youths, they'll be reflected in both adult and youth unemployment rates, and so we should still expect adult unemployment rates to predict youth unemployment rates. If youths were about a tenth of all trainees, that's not a lot higher than their proportion of the overall workforce (around 7% depending on the quarter you choose as labour force participation rates vary).