Friday, January 8, 2016

Why Retirees Go Broke

According to the American Bankruptcy Institute, the number of personal bankruptcy filings by Americans of all ages peaked at 1.5 million in 2010, the highest level since 2005, when the Bankruptcy Abuse Prevention and Consumer Protection Act made it more difficult to have debt forgiven. Filings declined to about 935,000 by 2014.

The Institute for Financial Literacy reports that older people are making up an increasing proportion of bankruptcy filers. The over-65 group made up 8.3 percent of all filers in 2009, or about 99,600, a rise from 7.8 percent in 2006.

Most bankruptcy filers were employed when they filed, but about 10% were retired.

Retirement research suggests that retirees can set up their spending and asset allocation to limit their “probability of ruin” to about 5% or so, but that is only the probability that he or she will deplete a savings portfolio as a result of market volatility and sequence of returns risk. Actually, there are several reasons a retiree might go broke.

The top five reasons for filing bankruptcy, according to a study entitled, “The (Interconnected) Reasons Elder Americans File Consumer Bankruptcy”, conducted by Dr. Deborah Thorne in 2010, are shown in the following chart recreated from her paper:

Other reasons for bankruptcies that were cited by filers and reported by the Institute for Financial Literacy included:
  • Divorce (15.1%)
  • Birth or adoption of child (9.7%)
  • Death of family member (7.5%)
  • Retirement (6.7%)
  • Identity theft (1.9%)
Respondents could choose more than one reason, so the total exceeds 100%.

The “retirement” reason includes both unplanned and unwanted retirement (another form of unemployment), and bankrupt retirees who believed they had adequate financial resources to retire but discovered they did not.

It is conceivable that some number of the filers who cited “retirement” as a cause for their bankruptcy succumbed to sequence of returns risk, though that data is not directly available. However, probability of ruin models show that portfolios are rarely depleted in less than 15 to 20 years for reasonable withdrawal rates, or until a 65-year old retiree is 80 to 85 years old. Research shows that bankruptcy filings decline significantly beginning at age 65 and the bankruptcy filing rate for age 85 and older is negligible. About 40% of elder bankruptcies are filed between ages 65 and 74. The fact that portfolio ruin is much more likely at older ages after bankruptcy rates actually decline suggests that sequence risk is probably not a large portion of this 6.7% of bankruptcy filings.

In other words, most bankruptcy filers in the study were too young to have depleted their portfolios as a result of a poor sequence of market returns.

Note that there is no category of reports of bankruptcies due to market losses or sequence of returns risk, so if this reason for bankruptcy were cited by any filers it wasn't in the top ten. One might reasonably expect “Income Problems (41%)” to include loss of income from assets, but a closer read of Thorne (2010) shows this category refers to unemployment issues and not loss of income generated from savings.

According to Thorne (2010), the probability of an American over age 65 filing bankruptcy is less than half a percent (about 0.43%), an order of magnitude less than the probability of ruin studies predict for premature portfolio depletion. The causes of bankruptcy are predominantly an unexpected increase in expenses, an unexpected loss of income or, more often, a combination of the two.

I will refer to the risk of bankruptcy from either lost income or unexpected expenses as spending risk, combining the two because the net result is the same whether we have too much expense or too little income, and a crises will often include both. Most retirement income research doesn’t address either disruptions due to large unexpected expenses or those due to loss of income, focusing primarily instead on the risk of outliving one’s savings resulting from disappointing market returns and poor sequences of returns. I will refer to the latter as earnings risk, or the risk that portfolio returns don't ultimately support the chosen spending rate.

Dr. Thorne goes out of her way to note that the reasons elder Americans file consumer bankruptcy are interconnected, including the term parenthetically in the study’s title. As I argued in two recent blog posts, Positive Feedback Loops: The Other Roads to Ruin and Retirement Income and Chaos Theory, I believe the causes are more than simply interconnected.

“I agree,” Dr. Thorne responded in an e-mail. “It's a cascade effect of really unfortunate events.”

Five top causes of elder bankruptcy: credit cards, illness, income problems, aggressive debt collection, housing problems.
[Tweet this]

Thorne notes in her study that “it appears that there is seldom a single reason for their bankruptcies; instead, elder debtors often file because of the cascading effects of multiple interrelated life crises, each as consequential as the last.” The percentage of respondents reporting the number of the five leading reasons for bankruptcy were:
  • None of the five reasons (8%)
  • One of the five reasons (22%)
  • Two of the five reasons (23%)
  • Three of the five reasons (27%)
  • Four of the five reasons (18%)
  • All five reasons (3%)
This is consistent with my theory that bankruptcies primarily result from positive feedback loops that initially develop from an income or expense shock, then form a positive feedback loop and spiral out of control. In other words, I believe that retirement income/expense systems, especially when spending crises are included in the model, are chaotic.

When credit cards are mentioned as a reason for bankruptcy, some assume that these households used consumer credit to live beyond their means. This is surely true of some households, but in many instances credit cards are the last resort for households whose finances are spiraling downward for other reasons, such as unemployment, medical expenses, or the death of a spouse or divorce.

For example, I helped a household with their finances in 2008 when bankruptcy was imminent. They owed more than $60,000 in credit card debt but the charges had been made for living expenses such as feeding and clothing three teenagers, not for shopping at Neiman Marcus. The reason for their bankruptcy was a prolonged period of unemployment.

We may have created the impression that retirees who invest in stocks and bonds and spend from a volatile portfolio have about a 5% probability of going broke, but retirees don’t go broke as a result of sequence of returns risk. They go broke as a result of illness, injury, unemployment, housing problems, divorce, birth or adoption, death or illness of a family member, forced retirement, identity theft, consumer debt and aggressive debt collection and the interconnected, cascading effects of all of the above.

Retirees can, however, lose their standard of living due to sequence of returns risk. This might or might not contribute to bankruptcy.

Are there actual retirees who go broke due to a sequence of poor returns? I’m not convinced. I’ve never met one. Or, read about one by name. I would think that if 5% of retirees were going broke for that reason, we would spot one or two occasionally and the elder bankruptcy rate would be much higher than half a percent. I have not found data describing the number of retirees whose standard of living was significantly lowered by sequence of returns risk, but I have seen anecdotal evidence from retirees who experienced this.

On the other hand, there were nearly 150,000 Americans over age 65 who filed for bankruptcy for other reasons in 2010.

Retirees who invest in equities are exposed to earnings risk, often referred to as sequence of returns risk or probability of ruin. All retirees are exposed to the risk of a spending crisis, whether or not they invest in a volatile portfolio. These are two very different risks.

Sequence of returns risk develops slowly and allows time for mitigation through spending reductions, requiring at least one to two decades to deplete savings. It might contribute to bankruptcy, but it is more likely to reduce the retiree's standard of living at worst. (Rational retirees will reduce spending when their savings decline in an effort to avoid ruin.)

Sequence of returns risk can be mitigated by reducing spending and, to a lesser degree, by managing portfolio allocation. This risk, which appears to be roughly 5% to 10% with reasonable withdrawal rates, is an order of magnitude more likely than the risk of bankruptcy from spending crises, but the magnitude of the risk is smaller, entailing reduced standard of living but probably not bankruptcy. This is the risk that attracts the most retirement research and planner attention.

Earnings risk can also be mitigated by a floor of safe, income-generating assets like TIPS bond ladders, annuities and Social Security benefits.

Spending risk, on the other hand, is a bolt of lightning that can reduce an apparently stable household to bankruptcy in a year or less. (I provided examples in Positive Feedback Loops: The Other Roads to Ruin.) Retirees who invest in stocks and those who don't appear about equally at risk of a spending crisis.

Once the downward spiral begins, it often cannot be stopped. Reducing spending is, by definition of the crisis, not an option – if we had the ability to adequately reduce spending, we wouldn’t be in a spending crisis. The magnitude of this risk is greater than that of earnings risk, entailing both loss of standard of living and bankruptcy.

Retirees with a volatile portfolio and a low, “safe” spending rate are not immune from spending risk. The low spending rate mitigates only the risk of portfolio depletion resulting from market volatility. It does not mitigate the risk of portfolio depletion resulting from say, a huge medical bill.

Spending risk can also be mitigated by a "floor" of protected assets, such as Social Security benefits held in a separate account or assets held in a retirement account. Delayed Social Security benefits would also be protected until received (the creditor risk is to benefits already received and commingled).

Spending risk is seldom considered in retirement studies. The closest relevant research is bankruptcy studies that allow us to separate data for older bankruptcy filers, though age is not an exact proxy for retirement status.

When we ignore expense and income shocks in our retirement models and simply assume that we will always be able to reduce spending whenever our portfolio balance declines, we ignore the risk of unacceptable outcomes from spending crises. Retirement plans should anticipate and plan for both risks. As Michael Kitces recently pointed out, a projection of future asset values is not a plan.

Our goal isn’t to avoid going broke due to market volatility and sequence risk.

It’s to avoid going broke.

Our goal isn’t to avoid going broke in retirement due to market volatility and sequence risk. It’s to avoid going broke.
[Tweet this]

A version of this post was recently published at Advisor Perspectives.


  1. Thank you for another thought provoking article.

    As a practical matter, how would a researcher isolate for sequence of returns risk as a contributing factor to bankruptcy?

    1. Good question, and not one that's easy to answer. Identifying any research strategy is a complicated process, but here are my immediate thoughts.

      First, SOR risk would ultimately show up as a bankruptcy due to depleted retirement savings, since none of the other categories seem to fit. As I noted in the post, there is no category that looks much like prematurely depleted savings except perhaps for the "Retirement" explanation. Even if the entire category resulted from sequence risk, we would still be looking at only 6.7% of bankruptcies. It seems more likely that a large portion of this 6.7% is the result of forced early retirement. Keep in mind that a retiree can deplete savings without going bankrupt.

      The second thing I would look at (did look at) is age at bankruptcy. Sequence of returns risks takes a decade or two to deplete a portfolio at a minimum. Bankruptcies before age 80 are highly unlikely to be the result of SOR risk and, as the data shows, bankruptcies decline dramatically beginning around that age.

      If I could identify the bankruptcies that occurred after age 80 and the claimed explanation were "Retirement", I could then look at market returns from the year of retirement until the bankruptcy and observe whether or not there had been an unfavorable sequence of returns during that period.

      So, I believe I could do so from the data used for Dr. Thorne's study and historical market returns, but as you can see from the process, it is highly unlikely that we would find many(any?).

      Excellent question. Thanks, Paul!