In which I argue that a real life “CaseBank” helps to protect against unforeseen and unintended consequences. And gives better precision in your evaluation of likely outcomes.

First, if you could please read this vignette .. it comes from Alan Kohler, a finance journalist and is about the consequences for an individual of a recent government policy change.

Yes, the example is about public policy, but I will talk about business shortly.

The most immediate and most appalling problem with the reforms is that the good things — tax on end-benefits and reasonable benefit limits (RBLs) both abolished — have been withheld for three months for “consultation” while the nasty, the imposition of new limits on contributions, applied from budget day. I can understand why Treasury would do that — to forestall a stampede to slide under the closing boom — but it is, in its effect, an absolute shocker.

To illustrate, here is the real life story of Jocelyn (not her real name), which is, unhappily, all too common. She retired last month at 67 having worked all her life running her own dress shop. She only had $152,000 in super because of the demands of running stock and debtors in the business. She simply could never pull enough out of the business to do much more than pay the mortgage and school fees….

Anyway, Jocelyn has now sold the business for $750,0000, most of which is stock and debtors, and has also sold a small investment property for $250,000 net of debt.

Up to Tuesday she had been planning to put that $1 million into super and pay herself a pension; but she can’t do that any more because, from budget night, a limit of $150,000 was imposed on undeducted super contributions, which is what it would have been.

This decision to announce a proposal to abolish tax on tax-paid retirement benefits and RBLs, subject to a three-month period of consultation, but at the same time to impose the offsetting clamp on the size of super contributions without any warning is causing chaos for thousands of people, like Jocelyn, who are in the middle of organising their retirement.

Intended, or unintended ? an oversight, or a known and accepted anomaly?

If Jocelyn had been in the CaseBank it might not have happened.

the CaseBank concept

A CaseBank is simply a collection of REAL LIFE “case studies” (more accurately, measurements of peoples situational and personal attributes), against which we evaluate the proposed change. NOT the artificial ones that you see in the explanatory literature (eg for tax laws) or the marketing literature (for studies of small business), but data about real people. Built – the data collected – BEFORE the mooted policy change, not in response to it.

Why real-life, and why prospective (not post-hoc) ?

Because we simply need to allow for the unusual, the range of things that ‘happen’ and not restrict it to our well-intentioned but undoubtedly restricted (and possibly self serving) vision of that range.

Without techniques (eg experimental designs) people simply are not good at thinking up (enough, and different) “representative scenarios”.

So, we turn to real life to provide them for us. And use our prior knowledge to stratify the data collection exercise so that we range as broadly as possible.

How large does a CaseBank need to be?

64 seems like a good answer .. that is not actually a totally flippant answer, it is just that with 64 data points you can control for quite a large number of factors.

So 64 would do, at a pinch.

But if the method you are using for evaluation (of the impacts of the proposed action) is not too expensive, or if you want detailed analysis of the impacts (perhaps by region or demographic sub group) then more data is better.

Marketers can perhaps relate to the CaseBank as a form of micro segmentation.

At one extreme we have “mass marketing” in which we consider what we propose to do against the backdrop of this thing called “the market” – one data point. Fine tuning this a bit, we might have “market segments” or “niches”, perhaps 6 segments each of which is somewhat internally homogenous (in characteristics, and in supposed response to the action) and each of which is quite different to the others. If we evaluate our action against each of the segments, then we have 6 data points : so, we expect some increase in precision/decrease in uncertainty about the likely outcomes.

If we have a nicely structured and balanced CaseBank of 64, we do 64 evaluations .. and get

• · a much better idea of the distribution of the impacts, and
• · a far better chance of identifying anomalies, and unintended consequences.

The construction of a customized CaseBank, the administration or replication or subsetting of the evaluations, the analysis .. those are topics for another day.