From Congress, a Law [JOBS Act] Befitting a Sausage Factory

Before decimilisation the minimum spread would be 1/16 (or even 1/8th) of a dollar. Today it is not uncommon to witness penny (or even fraction of a penny) spreads. Thus, the minimum theoretical spread realisable from making markets in a security dropped by 84% ($0.0625 to $0.01).

If previously broker-dealer A had a cost of one cent per trade and broker-dealer B had a cost of half a cent per trade, it just meant that B-D B made more profit - he couldn't undercut A. Thus both would survive. Today competition has driven the bid-ask spread to cost delta so low that market makers need large volumes to break even on their operations - this kind of volume can only be realised in frequently traded securities, i.e. larger companies.

Essentially, the wider spread subsidised inefficient market making operations, e.g. the making of markets in smaller companies. The positive externalities of those activities may have outweighed the dead-weight loss of the implicit subsidy.

Note that there is a valid counter-argument that says fractional pricing meant a company that today has a penny spread and a company that today has a dime spread would be seen as equally profitable to deal in, on a unitary basis, from the market maker's perspective, while today the less liquid stock is far more profitable (again, on a unitary basis). Thus, by this line of logic, decimilisation increased the incentive for market makers to make markets in less liquid securities, e.g. smaller companies.