If you see that a bottle of Chardonnay has dropped in price from $7.00 to $6.99, does it feel like you’re getting a bargain? The chances are it does – and you decide to buy it. In fact, the product is only one cent cheaper – so not such a great deal, after all! You probably wouldn’t buy it if it had fallen from $7.08 to $7.07. It is all because the left-hand digit [7] has been reduced.

This “left-digit effect” is an integral part of the study carried out by researchers Utpal Bhattacharya, Craig W. Holden and Stacey Jacobsen that looked at buy-sell imbalances on and around round numbers (7, as opposed to 7.28, for example), also known as “integers” Their focus was not the supermarket shelf, however, but the financial markets. As they point out, in an ideal world, those who initiate a trade would be equally likely to buy or sell at any given price point. In the real world, however, they often focus on round-number thresholds as reference points for value. If security traders focus on integers as reference points, a path that reaches or crosses a round number may generate waves of buying or selling. So, as in our example, a drop from $7.00 to $6.99 triggers more buys than a change from $7.08 to $7.07.

Two other round-number effects are highlighted in the paper: the “threshold trigger effect”, and the “cluster undercutting effect”. In the former, the idea is that investors have a preference for round numbers, where the hierarchy of “roundness”, from the most round to the least round, is (in the US) whole dollars, half dollars, quarters, dimes, nickels and pennies. So when the price reaches the integer it triggers trades. Limit-order clustering occurs when limit order (a direction to buy or sell at a specified price or better) prices are found more frequently on round numbers. The undercutting occurs when a new sell or buy is submitted at a penny lower or higher than the existing ask or bid. Cluster undercutting is a combination of these two effects.

To provide evidence for or against these three phenomena, the research team examined all trades of 100 randomly selected firms each year from 2001 to 2006; they had a sample of 137 million trades.

The researchers found excess buying by traders at all price points one penny below round numbers, and excess selling at all price points one penny above round numbers. They also discovered that the size of the buy-sell imbalance is monotonically ordered by the roundness of the adjacent round number – the largest imbalance above and below round numbers, second-largest above and below half-dollars, and so on. This supports the cluster undercutting effect.

Strong excess buying was seen when the ask price falls to reach the round number – similarly selling when the bid rises to reach the round number. There was relatively little buy-sell imbalance when the ask falls (or bid rises) to cross the round number. This supports the left-digit effect and threshold trigger effect.

The researchers suggest areas for further investigation. Liquidity-supplying, limit-order submitters might consider fighting their behavioral tendency to cluster on round numbers. It appears that cluster undercutting is a relatively profitable strategy that might be an improvement over clustering. Similarly, liquidity-demanding value traders might consider fighting their tendency to buy below (or sell above) round numbers. This could be done by intentionally switching their trading strategies to non-round price thresholds for action.