The Impact of Public Health Spending on the Rate of Gonorrhea: Evidence from California Counties Given the time constraint for your presentation, (i) don’t create too many (or too little) slides and (ii) don’t worry about discussing your paper in every detail. Hit the highlights in your presentation. Craig Gallet
I. INTRODUCTION While many studies have assessed the impact of public health spending on a variety of health outcomes, little attention has been given to the effect of such spending on the incidence of sexually transmitted disease (STD). Accordingly, this paper seeks to fill this gap in the literature. Utilizing panel data across California counties, we find increases in per capita spending on public health significantly reduce the rate of gonorrhea. Indeed, results from double-log specifications reveal that a 10% increase in per capita spending on public health reduces the rate of gonorrhea by approximately X%. Further, results are largely robust across various specifications. In the introduction, briefly motivate the topic you are studying. With respect to public health spending and gonorrhea rates, you could discuss increased rates in the presence of decreased spending on public health. See the answer key to Assignment 4 to ascertain the value of X in this slide.
II. LITERATURE SUMMARY Aggregate healthcare spending and health outcomes: Grossman (1972) Crémieux, P., Ouellette, P., and Pilon, C. (1999) Aggregate healthcare spending and STD: Chesson et al. (2005) Gallet (2009) County public health spending and health outcomes: Brown (2014) Briefly discuss the literature here. For instance, with respect to the impact of healthcare spending on health outcomes, you could mention studies that estimate “health production functions”. Only discuss a few papers, though, as you have a very limited amount of time to present your paper. From Brown (2014), since some public health spending is used to screen and treat STD, there is an argument for spending affecting the rate of gonorrhea.
III. EMPIRICAL MODEL AND DATA We estimate several health production functions using annual California county-level data for the 2003-12 period. Specifically: Linear Specifications: GONO = β0 + β1PUBHLTH + ε GONO = β0 + β1PUBHLTH + β2PHYS + β3P65 + β4PDEN + β5PUBLICU + ε GONO = β0 + β1PUBHLTH + β2PHYS + β3P65 + β4PDEN + β5PUBLICU + β6REGION1 + β7REGION2 + β8REGION3 + β9REGION4 + β10REGION5 + ε Double-Log Specifications: LGONO = β0 + β1LPUBHLTH + ε LGONO = β0 + β1LPUBHLTH + β2LPHYS + β3LP65 + β4LPDEN + β5PUBLICU + ε LGONO = β0 + β1LPUBHLTH + β2LPHYS + β3LP65 + β4LPDEN + β5PUBLICU + β6REGION1 + β7REGION2 + β8REGION3 + β9REGION4 + β10REGION5 + ε Discuss the various specifications you estimate. Provide your expectations and logic of how each variable affects the dependent variable.
Table 1. Variable Names and Definitions Variables Definition Dependent Variable: GONO Independent Variables: PUBHLTH PHYS P65 PDEN PUBLICU REGION1 REGION2 REGION3 REGION4 REGION5 REGION6 Gonorrhea rate (i.e., number of cases per 100,000 population) Real per capita spending on public health Number of physicians per 1,000 population Percent of the population age 65 and older Population density (i.e., population per square mile) = 1 if a CSU or UC campus is located in county i at time t, 0 if not = 1 if county i is located in San Francisco Bay Region, 0 if not = 1 if county i is located in Southern California (excluding LA) Region, 0 if not = 1 if county i is located in Los Angeles Region, 0 if not = 1 if county i is located in Central/Southern Farm Region, 0 if not = 1 if county i is located in North/Mountain Region, 0 if not = 1 if county i is located in Central Valley Region, 0 if not Sources: Data on GONO came from the California Department of Public Health (http://www.cdph.ca.gov) Data on PUBHLTH came from the California State Controller’s Office (http://www.sco.ca.gov/index.html). Data on PHYS and PDEN came from Rand California (http://ca.rand.org/stats/statistics.html). Data on P65 came from CDC Wonder (http://wonder.cdc.gov/). Region designations were based on a report from the California Department of Social Services (2002). PUBLICU, REGION1, REGION2, REGION3, REGION4, REGION5, and REGION6 were constructed by the author. Provide the data sources and define the variables.
Table 2. Descriptive Statistics Variables Mean Standard Deviation Minimum Maximum Dependent Variable: GONO Independent Variables: PUBHLTH PHYS P65 PDEN PUBLICU REGION1 REGION2 REGION3 REGION4 REGION5 REGION6 W X Y Z Briefly discuss some of the descriptive statistics. Perhaps counties and/or years in which GONO is highest/lowest, or counties/years in which PUBHLTH is highest/lowest. See the answer key to Assignment 2 for the values which go in this table.
Specification Exploration: The variable of interest in our model is spending on public health. With this in mind, we explore how sensitive the results are to various specifications. That is, we consider how the omission of other variables affects the impact of public health spending on the gonorrhea rate. Further, we consider the influence of alternative functional forms on the results. Regarding functional form, the relationship between the gonorrhea rate and spending on public health would be non-linear if there are diminishing returns to public health spending. Below is a plot of GONO against PUBHLTH which suggests non-linearity (see plot from Assignment 4): Your empirical model should explore the influence of specification choices. See answer key for Assignment 4 regarding plot of GONO against PUBHLTH.
IV. ESTIMATION RESULTS A. Linear Results Table 3. Estimation Results for Linear Specifications Variables (1) (2) (3) Constant PUBHLTH PHYS P65 PDEN PUBLICU REGION1 REGION2 REGION3 REGION4 REGION5 A (B) (B) # obs R2 Adj. R2 C D E A. Linear Results Note: Standard errors are in parentheses below coefficient estimates. *** Significant at the 1% level. ** Significant at the 5% level. * Significant at the 10% level. Discuss the estimation results in this slide and the next. See answer key to Assignments 2 and 4 for the numbers which should go in this table. Pay attention to the signs of the coefficients (as they relate to your expectations), significance of the coefficients, as well as measures of fit and any other tests performed (e.g., tests of multicollinearity, serial correlation and heteroskedasticity, as well as tests of the joint significance of various combinations of variables). Note there is a good chance your standard errors will need to be “robust”.
Note: Standard errors are in parentheses below coefficient estimates. Table 4. Estimation Results for Double-Log Specifications Variables (1) (2) (3) Constant LPUBHLTH LPHYS LP65 LPDEN PUBLICU REGION1 REGION2 REGION3 REGION4 REGION5 A (B) (B) # obs R2 Adj. R2 C D E B. Double-Log Results Note: Standard errors are in parentheses below coefficient estimates. *** Significant at the 1% level. ** Significant at the 5% level. * Significant at the 10% level. See the answer key to Assignment 4 for the numbers which should go in this table. Although I didn’t throw in the results here, as we explored in Assignment 4, you could also consider discussing the results in which we allow the impact of public health spending on the rate of gonorrhea to vary across regions.
V. CONCLUSION Summary of results Comparing results to the literature Policy implications Limitations and recommendations for further study Summarize your results here. Perhaps mention how they compare to other studies. Further, discuss any policy implications, as well as limitations and recommendations for future study.