The coefficient of primary interest is that associated with RANDR. Recall from Section 3 that for the dispersion of prices to decrease with increased information, one of two conditions had to hold. Either parents who value quality highly are wealdy more inclined to keep searching until they obtain a good match than are parents who have lower valuations of quality or parents who value quality in child care less must be more inclined to either drop out of the market or stay at the first provider they encounter.
Similarly, reduced form equations for maximum price and average price are specified as:
where MAXP^, is the maximum price for care in age group p in market m, and AVGPpm is the average price for care in age group p in market m. Note that under our model, better information will only result in lowered maximum and average prices if parents who highly value quality care for their children are sufficiently more willing to search for a good match than are parents with lower valuations for child care quality.
We separately estimate parameters for each of the three dependent variables, for each age group, using ordinary least squares. Standard errors are heteroskedasticity robust. Results are presented in Tables 4, 5, and 6 for the coefficient of variation, maximum price, and average price, respectively. Adjusted R-squareds range from 0.141 to 0.719 in the coefficient of variation models, from 0.234 to 0.761 in the maximum price models, and from 0.186 to 0.634 in the average price model.