As laid out in an earlier post, there are many ways to measure income volatility. The current diversity of methods, while frustrating to researchers seeking one, elegant statistical device, is actually a strength, providing a textured, fuller picture of Americans’ financial lives.
Another reason researchers should be wary of a universal method for calculating volatility is that the field is still grappling with why we should care about volatility, which will inevitably influence how we measure it.
For example, most agree that predictable volatility is not as problematic for families as unexpected swings. If you know weeks or months ahead that your wages will drop or your public benefits will expire, while you may not be happy, you will at least be able to plan accordingly. This could mean seeking a second job or saving more than normal during upswings.
The importance of predictability is one reason why administrators of the Federal Reserve’s Survey on Household Economics and Decisionmaking (SHED) are currently working on a new question on the predictability of work hours and wages. The U.S. Financial Diaries (USFD) made a point of asking families in their sample how predictable their monthly income is, with 33% reporting that their income was not easy, difficult, or very difficult to predict.
Though there are surely other ways to learn this information (e.g., by asking individuals to predict their income and then check their estimates against actual paystubs), a survey question is likely best since ultimately we care more about how individuals are perceiving – and planning around – their financial situation, than we do about how good they are at predicting the future. As economist Chris Carroll wrote in an answer to one of EPIC’s expert surveys, one key question in need of further research is the “extent to which fluctuations in income are predictable and/or chosen versus unpredictable and/or involuntary.” Devising an objective way to measure such a subjective concept is one of the most daunting challenges facing income volatility researchers today.
Another issue to consider when trying to measure volatility is the difference between income volatility, expense volatility, and consumption volatility. Measures of income volatility don’t capture fluctuations in expenses like medical emergencies, car repairs, and school supplies, and yet these shocks are, at least practically speaking, equivalent to a drop in income.
That said, neither income nor expense volatility is particularly worrisome if families have slack in their budgets and access to liquidity tools like credit and insurance. But for households living on the financial edge, misaligned inflows and outflows can lead to real hardships like utility shutoffs, food scarcity, and evictions. This is what some scholars call consumption volatility – swings up and down in the consumption of utility-enhancing goods and services. Some consumption volatility is fine (e.g., a temporary upgrade in living standards while on vacation) but most of the time families want to be steadily consuming the necessaries like food, shelter, and clothing.
Income and expense volatility often threatens to disrupt families’ access to those vital commodities. But documenting this link between volatility and consumption – and, ultimately, material hardship – can be tenuous.
Researchers often use spending as a proxy for consumption, which is sensible – especially for daily needs like food – but imperfect.
Finally, should we care more about volatility if it’s getting worse? It’s possible that income volatility has always been a fact of life for millions of Americans – which does not mean it’s not a problem, but may reduce the urgency with which policymakers tackle it.
Certain measures of volatility – like the coefficient of variation – have been collected for years, and there is evidence that both annual and month-to-month volatility has increased in recent decades. EPIC recently polled over 80 experts in the field – researchers, government officials, business people, and consumer advocates, among others – and found broad agreement that month-to-month income volatility has increased somewhat or greatly in the last decade (90%) and will continue to increase over the next decade (83%). But new data sources, like those produced by the JP Morgan Chase Institute, cannot easily reconstruct historical records, and the USFD only looked at families over the course of one year, so can’t speak to multi-year trends.
Interestingly, the 2015 result to the SHED income volatility question was very similar to that found in 2013, belying the widely held view that volatility is growing more severe. Two data points does not a trend make, so we’ll have to wait patiently for future iterations of the Fed’s survey to shed (sorry!) more light on whether or not incomes are growing more volatile.
Measuring income volatility poses both a technical and philosophical challenge. On the technical side, researchers face tough questions around how to treat certain types of income, who to include in a family unit, and what mathematical concept to use to express the results, among many others. The philosophical questions – why do we care about volatility and how do we link emerging research in the field to other important outcomes and historical trends – are even more daunting. Our hope is that EPIC provides researchers the forum necessary to grapple with these pressing issues.
