Betterment strives to help customers reach Behavior Gap Zero—a way of saying that bad behavior doesn’t interfere with achieving optimal returns.¹
To help our customers reach that goal, we’re always seeking to understand their motivations and decisions, and how those actions can make an impact on their investment returns.
However, measuring that impact is a multifaceted task, so we are starting with one simple question: How do allocation changes, or changing the weight of stocks to bonds in a goal, affect returns? Under what circumstances are customers making allocation changes, and are customers improving or reducing their returns with these changes?
Key Finding: Market Timing Hurts Returns
After sifting through the data, we found these key points:
- In 78% of accounts, customers made less than one allocation change per year, consistent with occasional changes in financial circumstances.
- Across all accounts, 67% of customer accounts have Behavior Gap Zero.
- Account holders who are more active appear to make allocation changes in reaction to the market, and underperform as a result.
- We observe an average behavior gap of 22 basis points across all accounts; however, we noted a skewed distribution, and 74% of accounts are better than this average.
Below, we first detail the ways our smarter technology is making allocation selection easier than ever for investors. Second, we review the results of our behavioral research. Together, these illustrate the unique ability and dedication we have at Betterment to understanding and assisting with the behavioral components of investing.
What makes this data unique?
Most existing studies of the behavior gap in investing look at cash flows into and out of different funds. This is a necessity because most investing platforms require that an investor move funds from one vehicle to another to adjust risk.
Existing studies generally show that those cash flows mistime the market, on average: rushing in after periods of success and retreating when returns are down. But cash flows are a muddy measurement of risk preferences. Investors may move assets because of fees, manager reputation, to consolidate investments, or because they need the money to spend.
For investors, adjusting risk allocation at Betterment is as simple as dragging a slider to change the proportion of stocks and bonds within the portfolio. As a result, we can focus on how investors change allocations while filtering out other confounding changes, which allows us to measure the resulting impact on returns with great clarity.
Automated Allocation: A New Paradigm
We understand the reality of changing financial circumstances and accordingly, we make our allocation change as easy and efficient as possible.
Adjusting portfolio allocations has traditionally been full of frictions: phone calls to advisors, fees for trades, and uncertain tax consequences. Betterment customers experience none of those frictions, and as a result, have a new level of freedom and power to adjust their allocation to meet their financial needs.
Imagine a career change that pushes plans to buy a house back by three to five years, or a family change, such as a divorce or death, that makes retiring a few years later seem like the right decision. These changes in the time horizon of the investing goal mean that stock allocation should be increased accordingly.
On the other hand, if an investor needs the money sooner than originally expected, more bonds are appropriate to minimize uncertainty around the withdrawal date.
Betterment reduces certain costs and barriers to allocation changes (e.g., fees, fund choices, taxes) so that our customers always have the right portfolio for their financial situation. What we don’t want to encourage, however, are allocation changes in response to market movements or guesses about what the market will do in the future. As long-time Betterment customers know, we discourage market timing at every turn.
With this research, we wanted to answer two questions:
- Are customers trying to time the market?
- If so, how is that affecting their returns?
Examining Customer Behavior
The data below shows allocation changes as they coincide with market performance (dark blue line) during just the last two quarters of 2014. Each bar shows the average direction and magnitude of allocation changes over the previous seven days. Note that these averages include only the small number of customers who made a change over the 7-day period— typically five out of every 1,000.
When focused on the tiny subset of customers who were active, we do see some behavior that looks like market timing. How do we know this? If these investors were making allocation changes purely in response to idiosyncratic changes in their financial circumstances, the average change would stay close to 0% and we wouldn’t see any correlation with market performance.
Allocation Changes and Market Performance
Data Shows Subset of Active Users Only
We see that changes followed the market to some degree in the second half of 2014, but what about a more systematic trend?
If we use all of our data since 2010, we can look for such systematic patterns.² The plot below shows the proportion of allocation changes moving toward stocks or bonds based on market performance in the preceding week.
We see that when the market is up between 0% and 3% in the preceding week, the average change increases the stock allocation by 6%. When the market is down more than 2% in the preceding week, the average allocation change is toward bonds by 5%. Remember, more than 99% of customers do not make an allocation change during a given 7-day period, so this pertains to the decisions of the small number who do make changes.
Proportion of Allocation Changes Moving Toward Stocks or Bonds Based on Market Performance
How is this affecting returns?
Perhaps these active customers are actually improving their returns by making changes? To measure the effect of allocation changes on a customer’s returns, we compared the returns the customer experienced as a result of actual allocation choices to an individualized benchmark.
For the benchmark, we used the average time-weighted allocation chosen by the customer. Then, we compared each customer’s actual returns resulting from actual allocation levels over time to the benchmark: the returns that would have resulted if the customer had been at the average allocation constantly from day one. In both cases, we used time-weighted returns, so cash flows into or out of the account did not have an effect.
As an example, imagine a customer who funds a five-year home-buying goal and sets the allocation to 70% stocks. After one year, the customer increases the allocation to 95% to chase the performance of surging U.S. equity markets. A more volatile second year makes the prospect of losses more salient, and the customer sets the allocation back to 70%. By the end of the third year, the account had earned 18%. In this analysis, we compare that actual 18% time-weighted return to a hypothetical account with a constant 78.33% allocation over the same period (the time-weighted average of 70, 95, and 70).
By doing this analysis on every account, we find making allocation changes reduces returns, on average.
By doing this analysis on every account, we find making allocation changes reduces returns, on average. Across all Betterment accounts, the mean gap between actual returns and the average allocation returns was 22 basis points. However this includes all the accounts that made no allocation changes, and whose gap would by definition be zero. Looking only at the accounts that made at least one allocation change, the average gap is 41 basis points.
The figure below shows that each additional allocation change (up to four) increases the average behavior gap by about 16 basis points. Only 22% of accounts have made more than one allocation change per year. Nearly half have never made even one, so there is no possibility of a behavior gap, as we define it here. Overall, two-thirds of customers have no gap, and three-fourths have a behavior gap smaller than the 22 basis point average.
Behavior Gap and Allocation Changes
IRAs Are Different
Interestingly, we see customers making fewer allocation changes in their retirement goals in IRA accounts, and their returns are closer to the ideal as a result. Betterment customers make 48% fewer allocation changes in their IRA accounts and see 50% smaller behavior gaps, on average. We see two potential explanations for this difference.
First, it’s likely that our customers’ retirement horizons don’t change very often, so there are fewer legitimate allocation changes than in many of the shorter-term taxable savings goals. Second, the findings are consistent with prior research showing less activity in retirement accounts, possibly because they are perceived as more “off limits.”³
It’s worth noting that a preference for changing allocation inside a taxable account versus an IRA, all else being equal, is highly tax-inefficient behavior, as investors are not taxed on realized gains inside an IRA. The returns analyzed here are not after-tax, so the findings actually understate the gap that results from frequent trading in a taxable account, which we’ve written about here. If an investor is going to market time (and we advise against that!), doing so in a taxable account is particularly ill-advised, as compared to an IRA.
How does this apply to individuals?
Does this mean that an investor should never make an allocation change? Not at all. If an investor’s financial goals or constraints have changed, an allocation change may be the right thing to do. In that case, we recommend going to the Advice tab in your Betterment account, entering the new time horizon or target balance, and adjusting accordingly. Using allocation changes as a tool to try to time the market, or in reaction to periods of bullish or volatile markets, is where we see trouble. These kinds of changes only make sense if an investor know exactly what the markets will do next, and that is not something most of us should be betting on.
¹ The “behavior gap” is a term coined by financial planner and journalist Carl Richards and has become a popular way to refer to the loss of investor returns due to bad timing decisions.
² While the average market movement has been upward over the period Betterment has been available to investors, it has been marked by drawdowns of up to 20% in the S&P 500 since 2010.
³ Barber, B. M., & Odean, T. (2000). Trading is hazardous to your wealth: The common stock investment performance of individual investors. The Journal of Finance, 55(2), 773-806.