What a Narrower Spread Really Tells You

The evolving U.S. corporate-bond market is the subject of this final synthesis, and the question is what a falling bid-ask spread allows us to conclude about it. The evidence supports a precise answer: a narrower spread can mean that a customer is paying less to complete an immediate trade, while saying little by itself about dealer capacity, the compensation investors require to hold the bond, or the market's ability to absorb a wave of selling. The most useful judgment is therefore not whether the market is simply “liquid,” but which form of liquidity improved and which burden may have moved elsewhere.

The combined result is a separation, not a contradiction

The Philadelphia Fed paper reaches its clearest conclusion by changing three features of the dealer-intermediated market together. The counterfactual raises balance-sheet costs by 32 basis points, increases the electronic share of trading from 20% to 50%, and raises turnover from 11% to 16%. A basis point is one-hundredth of a percentage point; the important comparison is not the unit alone but the stated change from one market condition to another.

Table 2 reports the isolated and combined effects as follows:

The combined row is the answer to the opening question within the model. The liquidity yield spread rises by 0.87 basis points, or about 2.8%, while the average bid-ask spread falls by 6.29 basis points for voice trades and 4.41 basis points for electronic trades. The corresponding percentage declines in the two bid-ask measures are about 37.7% and 37.2%. In other words, customers appear to face much lower per-trade transaction costs, yet the model's liquidity component of the cost of capital is almost unchanged and slightly higher.

There is no logical inconsistency because the three changes act on different margins. Electronic RFQs make customer search more competitive and directly reduce the surplus a dealer can retain from a feasible quote. Higher turnover makes customers need to trade more frequently, raising the value of reliable access and the compensation required to hold the bond. Higher balance-sheet costs make inventory more expensive, reducing the number of dealers willing to carry it. The electronic improvement pushes the liquidity yield down; the demand and capacity changes push it up. The final yield is the result of those forces meeting in equilibrium.

The isolated effects show why it would be misleading to ask whether each change was “good” or “bad” for liquidity without naming the measure. The balance-sheet-cost experiment raises the liquidity yield by 5.43 basis points while producing only small, venue-dependent bid-ask movements. The electronic experiment lowers all three reported spread measures. The turnover experiment raises the liquidity yield by 6.47 basis points while leaving the voice and electronic bid-ask measures essentially flat. Each row is a different answer to a different liquidity question.

The combined result is also not just the sum of the rows. For the liquidity yield, the isolated effects sum to +3.90 basis points, but the interaction effect is −3.03 basis points, leaving +0.87 basis points combined. For voice bid-ask spreads, the isolated effects sum to −6.37 basis points and the interaction is +0.07, leaving −6.29. For electronic bid-ask spreads, the isolated sum is −3.72 and the interaction is −0.68, leaving −4.41. The interaction terms matter because changing one part of the market changes the conditions under which the other changes operate.