Since the PSA bonus was in effect from 2005 to 2008, we applied a DID estimation method to identify the effect of elimination of PSA bonus on the following outcome variables: number of eligible providers, practitioners’ volume of services, and their total payments. We used the provider level data on the number of E&M claims and allowed charges associated with those claims from 2005 to 2011. The data from 2005 to June, 2008 show the scenarios in the pre-elimination (of PSA bonus) period and the data from 2009 to 2010 provide outlook on the post-elimination period. Data from 2011 will not be used for the DID analysis as it will include the confounding effect of the Medicare PCIP bonus. The sample is restricted to physicians with the following PSA bonus eligible primary care specialties: family practice, internal medicine and OB-GYN.
Equation (1) below shows the econometric model we estimated to assess the impact of PSA bonus policy:
The model in equation (I) is specified based on the provider level data in each year. Here, the dependent variable, , is the volume of primary care (E&M) services provided by primary or non-primary care practitioner i at time t; is an indicator for the treatment group that takes a value of 1 if the provider is in a primary care PSA area and zero otherwise; is an indicator variable taking the value of 1 if the services are provided during 2005-2008 period (when PSA bonus program was in effect) and zero otherwise; the vector includes variables reflecting the provider characteristics, features of their geographic locations etc., and is a time-specific fixed effect term. The term represents random unobserved factors affecting services.
If elimination of the PSA bonus was associated with a reduction in the volume of primary care E&M services provided by practitioners in those areas, then we should expect the estimated value of to be positive. A positive estimated value of would imply higher volume of services during the period when the PSA bonus was in effect. In addition, the interpretation of the first four terms in equation (I) is as follows:
= the average number of E&M claims during the PSA period (PRE), for providers located in the PSA regions;
= the average number of E&M claims during the Post-PSA period, for providers located in the PSA regions;
= the difference in the average number of E&M claims between the PSA and Post-PSA period, for providers located in the PSA regions;
= the difference in the average number of E&M claims between the PSA and Post-PSA period, for providers regardless of location.
Exhibit 21 shows the results from the estimated DID model to identify the impact of the PSA bonus policy on the supply of primary care services. We found that on average providers with PSA bonus eligible specialties irrespective of location were estimated to have about 153 more claims submitted during the PSA period compared to the post-PSA period. However, bonus-eligible providers located in PSA areas had on average an additional 50 claims per year compared to those located in non-PSA areas. We attribute this additional increase in the volume of primary care E&M services to the PSA bonus policy. On the other hand, we do not find a statistically significant impact of the PSA bonus policy on the total annual allowed charges per physician (column 2).
Exhibit 21: Impact of PSA Bonus on the Number of
E&M Claims under Medicare, 2005-2010
|Number of E&M Claims(1)||Allowed Charges for E&M Claims ($)(2)|
|Location PSA PC Indicator (PSAPC)||70.55*||5622.0*|
|PSA Year Indicator (2005-2008)||152.6*||870.9*|
|PSAPC* PSA Year Ind.||49.84*||-241.2|
|Median Income ($10k)||-10.47*||926.7*|
|Percent under poverty||-6.671*||-352.7*|
|Percent Population over 65||19.36*||1389.4*|
|Primary Care Phy./mcarepop10k||-0.110*||-2.049*|
|Specialty Fixed Effects||Yes||Yes|
|Year Fixed Effects||Yes||Yes|
|State Fixed Effects||Yes||Yes|
Note: Robust standard errors are in parentheses; + significance at 10 percent; * significance at 5 percent. The sample is restricted to only providers with the following PSA eligible specialty: family practice, internal medicine and OB‑GYN.
Exhibit 22 shows the results from the estimated DID model to identify the impact of the PSA bonus policy on the number of primary care providers at the zip code level. As explained above, our primary focus is the estimated coefficient of the interaction term between the PSA status indicator of the location of physicians and the PSA period (i.e., 2005-2008). The number of providers with PSA bonus eligible specialties is estimated to increase by about one in PSA areas during the PSA period compared to the number of providers in non-PSA areas. This represents a meaningful increase given that, on average, there are about six primary care providers in PSA areas. We attribute this additional increase in the number of primary care providers to the PSA bonus policy.
Exhibit 22: Impact of PSA Bonus on the Number of Primary Care Providers per County (2005-2010)
|Number of Providers with PSA Bonus Eligible Specialty|
|Location PSA PC Indicator (PSAPC)||-4.951*|
|PSA Year Indicator (2005-2008)||-1.722*|
|PSAPC* PSA Year Ind.||1.149*|
|Median Income ($10k)||0.468*|
|Percent in poverty||0.256*|
|Percent Population over 65||0.139*|
|Primary Care Phy./mcarepop10k||0.146*|
|Year Fixed Effects||Yes|
|State Fixed Effects||Yes|
*Note: Robust standard errors are in parentheses; + significance at 10 percent; * significance at 5 percent. The sample is restricted to only providers with the following PSA eligible specialty: family practice, internal medicine and OB-GYN.