Remembering back to our last two posts where we looked at how two demographics either impacted upon the ALP vote at the last election, or correlated strongly with something(s) that did, we were very careful to compare like with like – either a stock (the ALP TPP vote) with a Stock (the percentage size of some population in an electorate), or a flow (the change in the ALP vote) with another flow (the change in the size of some population in the electorate).
It might be a good idea to highlight why this is important with an example based on the data we’ve been using over the last 2 days.
Yesterday we found that as the growth in the population aged 0-4 yrs increased in each electorate, on average the swing that the ALP received at the last election in each electorate decreased.
But what happens if instead of comparing a flow with a flow, we compare a stock with a stock flow (many apologies about that folks – I’m sure plenty of you were going ‘WTF is that furry critter on about’)? If we compare the ALP TPP vote by electorate (a stock) with the change in the size of the 0-4 yr age cohort by electorate (a flow) we end up with a chart that encourages misleading interpretations of reality.
This suggests that as the growth in the size of the rugrat population increased, so too did the ALP TPP vote – the exact opposite of what we discovered yesterday.
The reason we get misleading results is that stocks are not only a function of the most recent flow, but a function of all cumulative flows of all previous periods. We are effectively comparing here how one thing changed in one period of time (the change in the population of rugrats over 2006/07) against the aggregated results of how another thing changed through all previous periods of time. This gives us a huge intertemporal disconnect where we are comparing one year against the cumulative results of over 100 years.
If we run the above chart again, but this time we use like with like, flow with flow, we can see how things changed in a single period of time – which is what we need to look at if we are to extract real meaning from the observable reality of the single 2007 election. This chart tells the real story, and a very different story to the first chart.
Unfortunately, this like with like comparison gets ignored far too often – not just in the social sciences but with economics. You’ll often hear things like the ratio of debt to income being used to tell us how the sky will fall in because of our enormous debt burden. And while the sky may or may not fall in, that ratio wont have much to do with it.
There’s nothing wrong with that ratio, it’s a mighty fine ratio – if you’re doing some pretty obscure econometric analysis on a few niche subjects, or using it to measure a function of time (such as how the ratio of the median house price to the median household income has changed over time, which gives us an insight into how the length of time to pay off a mortgage for the median purchaser has changed through the years).
But as a meaningful statement on the size of our debt burden it’s horsefluff.
Debt is a stock, income is a flow –we need to compare like with like such as debt servicing to income, or debt to assets, but debt to income is a dodgy stat not worth the hot air it’s usually said with.
So keep these things in mind when you see charts and comparisons – the rule is simple; stocks with stocks and flows with flows unless there’s a profoundly important reason not to (and where hyperbole is never one of them).
2 Comments
Hi Possum
Nice post
I agree that one has to be very careful about comparing stocks with flows, and that discussions of the debt-income ratio can be quite misleading: a) because interest rates have fallen so much since the 1980s a larger stock of debt can be serviced with the same proportion of income; and b) debt may simply have risen in line with assets.
However, there are circumstances where discussing the debt-to-income ratio can still come in handy. First, debt has to be serviced out of current income. Thus a debt-to-income ratio convey information about how vulnerable the household sector is likely to be to a sudden increase in interest rates. Second, asset prices can be volatile, and as we know, subject to bubbles. Simply looking at leverage ratios can be misleading if increases in asset prices are not based on fundamentals. To my mind, if nominal credit is growing persistently faster than nominal income, it isn’t in of itself something to worry about. But at the same time, responsible policymakers do have a duty of undertaking analysis to determine whether there are good fundamental reasons for nominal credit growth to be growing more rapidly than nominal income, and what vulnerabilities might arise from it.
Another example of where comparing stocks to flows has some use is Australia’s external accounts. The interest on Australia’s net foreign liabilities have to be serviced out of current Australian income. An increase in net foreign liabilities as a share of GDP tells us that for a constant yield on liabilities, Australia’s net income deficit is likely to be higher, and consequently the CAD. Now, of course, there may again be fundamental reasons why an increase in the ratio of NFL to GDP is not a concern – but it suggests a vulnerability…
Anyway, thanks again for the informative post…