Australians pride themselves on having dubious origins. The Australian Institute of Family Studies (AIFS) fits that bill. Conceived as an afterthought to the Family Law Act of 1975, the Institute’s birthing took fully another 5 years, and another Act of Parliament to patch up botched initial legislation. Not only were the circumstances of its birth unfortunate. The Institute’s original mandate seemed to give it little scope for action. The Act charged it with (and I quote) studying “factors affecting marital and family stability in Australia”.¹ Taken very literally, and seen in a narrow family law context, that could have been pretty limiting.

But right from the start, the Institute saw the impact of work on family life as an important part of the “factors affecting marital and family stability”. It has been studying that for three decades now: through its path-breaking Maternity Leave Study; through the book Work and Family Life (Wolcott & Glezer, 1995), and the magisterial Australian Living Standards Study underlying it; and continuing into the present day through the Institute’s research theme on “Families and Work”.²

Among the many high-quality research tools available for exploring those issues in Australia, I want to draw attention to one in particular: the Australian Time Use Study. Piloted in 1987, the Australian Bureau of Statistics (ABS) has been conducting time use studies (increasingly intermittently, which is a great pity) for almost a quarter century now. Those ABS studies have become the gold standard, methodologically, among time use researchers worldwide (Stinson, 1999).

From the earliest days of those time use studies, family researchers have been exploring them to cast light on issues of work–life balance that confront Australian families. Janeen Baxter and Diane Gibson’s book, Double Take (1990), and Michael Bittman’s on Juggling Time (1992) stand alongside international classics such as Juliet Schor’s The Overworked American (1991) and Arlie Hoschshild’s The Second Shift (1989) and The Time Bind (1997) in that tradition.

One key insight runs through all those books. That is the observation that people can be “time poor” as well as “money poor”, and that those two may not always track one another closely.

Here’s one back-of-the-envelope way of getting a sense of that. Begin with the standard way of calculating poverty rates in terms of money—as half of the median equivalent income (i.e., household income adjusting for household size) across the country as a whole. Call anyone who has less income than that “money poor”. Then recall the international convention (to which Australia is a signatory) that says people should not have to work more than 40 hours per week in paid labour.³ Let’s take that as a rough-and-ready indicator of time poverty. Call anyone who spends more than 40 hours per week in paid labour “time poor”. Then just go to the Luxembourg Income Study database and run the numbers.

In one pretty typical year, the proportion of the prime working-aged Australian population who were “money poor” was just under 13%.⁴ But if you ask, “What proportion of the population was either time-poor or money-poor?”—that is, were poor by one or other (or both) of those standards—the answer jumps to a whopping 51%. Truly. Over half the prime working-aged Australian population either had an income below the poverty line and/or worked more than 40 hours a week earning that income.

Well, that’s certainly a finding that grabs our attention. But what exactly are we to make of it? Everything depends on what exactly is going on with the extra two-fifths of the population that is time- or money-poor, over and above those who are money-poor alone.

There are two possibilities, here. Maybe, in case 1, they were academics, working 50 hours a week for sheer love of their subject. In that case, we wouldn’t be very worried; they’re doing what they’re doing purely for fun. But maybe, in case 2, they have to work long hours just to push their incomes above the poverty line. That would worry us.

The point is well captured in a joke that was current in the US during the Clinton Administration:

Q: “Why are you complaining about the state of the economy? Don’t you know that President Clinton has created 2 million new jobs?”

A: “Yes, and I’m holding down three of them!”

Not good, if that’s the only way the person can stay out of poverty.

It is a great shame that ordinary time use studies fail to make this crucial distinction: between people who are working long hours because they have to, and people who are working long hours because they choose to. Ordinary time use studies simply count up how many hours a day people spend doing each of several things: how many hours they spend in paid labour; how many hours in unpaid household labour (shopping, cooking, cleaning, taking care of the kids); how many hours in personal care (eating, sleeping, personal hygiene). In ordinary time use studies, how much “spare time” people are said to have is just how much time they have left over after deducting time actually spent in those three broad classes of activities.

But that way of looking at things conflates the cases of the university researcher working long hours for fun and the person working three jobs just to get by. To pull those cases apart, what we require is a measure—not of how much time people actually spend in paid labour—but rather of how much time they need to spend in paid labour. The same applies for all those other sorts of activities I mentioned. (After all, lots of people spend more time than strictly necessary cooking for fun, keeping the house fastidiously tidy and lounging in bed, just as lots of people spend more time than the minimum necessary in paid labour.)

With my colleagues James Rice, Antti Parpo and Lina Eriksson, I developed and operationalised a concept of “discretionary time” that attempts to make just that distinction. We try to specify how much time people strictly need to spend in each of those sorts of activities. What’s left over we call their “discretionary time”. They may—they probably will —choose to spend some of their discretionary time doing more of those same things. That’s fine. There’s no reason they shouldn’t. It’s perfectly reasonable for people to want more than a minimally adequate income, more than a minimally adequate meal, more than a minimal night’s sleep. Our point is just that, when doing more than the minimum that is strictly necessary, that should be seen as a choice; that is, their chosen way of how to spend some of their discretionary time, not a limit on how much discretionary time they have available to spend.

Of course, the trick lies in how to define what is “minimally necessary” in all these realms. I’ll tell you how we did it in a minute; and I’ll defend that as being superior to alternative approaches to that task, whenever called upon to do so. But I want to emphasise, above all else, the importance of finding some way—whether ours or some other—of operationalising this crucial distinction between choice and necessity. Time use studies can seriously mislead us if that crucial distinction is not made—in ways that matter for public policy. (I’ll show how shortly.)

Here is how we operationalised “necessity” in each of the three major realms of time use. “Necessary time in paid labour” we define (unsurprisingly) in relation to the standard poverty line: half the median equivalent income in your country. How much time is necessary for you to spend in paid labour is, we say, however long it would take someone at your current wage rate to get an income over the poverty line. The poverty line is a measure of necessity; anything above that is discretionary.

That’s easy, because there is such a widely agreed measure of poverty in the realm of money. We piggyback on that in devising a measure of “necessary time in unpaid household labour”. By analogy, we say that how much time you need to spend on that is just half the amount of time spent on that by the median person in your country living in a household with the same structure as your own. (Gratifyingly, specifying necessity in this realm in this way, we find that about the same proportion of people fall below that standard of “necessity” for unpaid household labour as for paid labour. That doesn’t exactly “validate” that instrument, but it at least offers some welcome reassurance.)

Alas, we cannot implement exactly the same procedure when it comes to estimating “necessary time in personal care”. Merely taking half of what the median person does there would yield an unreasonably low estimate. (The reason for that is simple: “personal care” time consists very largely of sleeping and, although we all laze about from time to time, virtually no one gets literally twice as much sleep as he or she needs.) Instead, we take four-fifths of the amount of time the median person actually spends in personal care as our measure of what is necessary to be done in those activities. (That proportion was chosen because it yields roughly the same proportion of the population doing less than “necessary” as in the other two classes of activities.)

There are, as I say, other ways to specify what is “necessary” in each of those realms. I’d like to be able say, “They all come to roughly the same in the end”, but we don’t know that yet. No one else has systematically implemented any of those alternative strategies on any nationwide dataset that I know of. So all I can do is report how much of a difference it makes to distinguish, in the particular way that we have done between choice and necessity in the use of time.

First, let me describe how things would look if we were to follow the ordinary practice of time use researchers. Remember, they focus on how much time people actually spend in all those activities, and calculate “spare time” as what is left over. Well, by those standards a lone mother would be no more time-poor than a childless woman in a dual-income household (call her a “DINK” for short: dual-income, no kids). Calculating things the way time use researchers conventionally do, both DINKs and lone mothers would have had around 30 hours of “spare time” per week.⁵ But does anyone seriously believe that a lone mother is no more pressed for time than a dual-earner with no kids? That just can’t be right.

What’s gone wrong in the ordinary time use calculation of “spare time” is, once again, a simple failure to distinguish between choice and necessity. It’s perfectly true that, once you subtract out all the hours that the DINK, actually spends at her desk, she has no more “spare time” than the lone mother. But what those calculations miss is the fact that the lone mother has to spend much more of her time in the ways that she does, whereas with the DINK spending her time as she does is to a much larger extent her discretionary choice.

The magnitude of that difference is revealed by our calculations of “discretionary time”. Instead of focusing on how much time people actually spend in those activities, we ask, “How much time is it strictly necessary for them to spend in those activities?” How much discretionary time they have is simply how much is left over, after that is subtracted away.

When the calculations are performed in that way, the DINK and the lone mother appear as polar opposites. Lone mothers in Australia had around 60 hours of “discretionary time” per week, in contrast to 87 hours for DINK women. That’s to say, in terms of control over their time, lone mothers were 27 hours a week worse off than DINKs. Think of it: 27 hours. That’s like having to work three extra days a week (Goodin, Rice, Parpo, & Eriksson, 2008, chapter 5). That is a huge difference. But it only comes into view if we abandon the ordinary methodology of time use researchers and adopt instead some sort of methodology that is sensitive to the difference between choice and necessity in the use of one’s time.

At international conferences of time use researchers, I have entered pleas that we make this distinction. I am pleased to report that all the main players there are wholly onside; as well as the audience at the monthly seminar of the Research Division of the World Bank. So I hope, and believe, that we are getting some traction on the problem, and general practice will soon be reformed. (That is to say, we’ll soon get new measures to supplement the old—which remain, of course, useful for other sorts of purposes, I hasten to add.) But while I’m confident that we will eventually get it sorted out, in the meanwhile, family researchers need to know that there are problems with the data currently on the table and that makers of family policy might be seriously misled by it.

Rest assured. In terms of discretionary control over their time, lone mothers really are temporally the worst off people in the community, notwithstanding the false impression you might get from looking at time use data as it has traditionally been presented.