Fixed Number of Officers

Counting officers as full number of officers, not pruned number of officers (excluding those with limited history/future).

Fixed City Level HPI

Facts

Fact #1: There is considerable variation even in the pay of police officers – with a range from 45,575 to 206,821.

Person-Level Data

base_plus_overtime Base_Pay Overtime_Pay
n 8209 8209 8209
mean 115923 93919 22004
sd 32285 18326 22122
median 111031 92837 16098
min 2906 2906 0
max 450373 218508 322071
range 447467 215602 322071
skew 1 0 3
kurtosis 6 2 14
se 356 202 244
Q0.01 57219 47663 0
Q0.05 75422 67233 774
Q0.1 83133 72382 2556
Q0.25 95117 83952 7485
Q0.5 111031 92837 16098
Q0.75 130861 104793 28842
Q0.9 154229 117340 48062
Q0.95 174797 125868 64339
Q0.99 223608 135386 105578

Agency-Level Data

base_plus_overtime Base_Pay Overtime_Pay
n 218 218 218
mean 109526 89521 20006
sd 25519 19235 10831
median 109602 88813 17673
min 45575 43403 0
max 206821 158108 63034
range 161246 114705 63034
skew 0 0 1
kurtosis 1 0 1
se 1728 1303 734
Q0.01 58730 49320 1400
Q0.05 69822 59911 5459
Q0.1 76035 64470 8089
Q0.25 92344 76417 12977
Q0.5 109602 88813 17673
Q0.75 125101 101955 25488
Q0.9 143151 115549 34402
Q0.95 151824 120282 41475
Q0.99 169453 129585 52697

Fact #2: There is considerable correlation (coefficient = .39) between base pay and overtime pay

Agency-Level Data

Fact #3: Overtime is a larger share of compensation in places where total compensation is higher (this is partially mechanical)

Agency-Level Data

Fact #4: Big Cities Pay More

Agency-Level Data

Fact #5: Bigger Cities have more cops

Fact #6: Richer Places Pay More, and both size and wealth are significant if you income them both

## 
## ===============================================
##                         Dependent variable:    
##                     ---------------------------
##                         base_plus_overtime     
## -----------------------------------------------
## log_pop                    17,146.000***       
##                             (2,587.000)        
##                                                
## log_income                 75,370.000***       
##                             (7,771.000)        
##                                                
## Constant                  -332,329.000***      
##                            (39,396.000)        
##                                                
## -----------------------------------------------
## Observations                    218            
## R2                             0.389           
## Adjusted R2                    0.384           
## Residual Std. Error    20,032.000 (df = 215)   
## ===============================================
## Note:               *p<0.1; **p<0.05; ***p<0.01

Fact #7: There is no correlation between income and number of police officers once you control for city level population

## 
## ===============================================
##                         Dependent variable:    
##                     ---------------------------
##                            log_officers        
## -----------------------------------------------
## log_pop                      0.808***          
##                               (0.035)          
##                                                
## log_income                     0.048           
##                               (0.104)          
##                                                
## Constant                     -2.520***         
##                               (0.527)          
##                                                
## -----------------------------------------------
## Observations                    218            
## R2                             0.717           
## Adjusted R2                    0.715           
## Residual Std. Error      0.268 (df = 215)      
## ===============================================
## Note:               *p<0.1; **p<0.05; ***p<0.01

Fact #8: The Correlation between Pay and Crime Rates is negligible for violent crime – and there are lots of places with almost no crime that are paying cops a lot

Fact #9: Both violent and property crime predict the total number of police officers

## 
## =====================================================
##                            Dependent variable:       
##                     ---------------------------------
##                               log_officers           
##                           (1)              (2)       
## -----------------------------------------------------
## log_prop_crime          0.675***                     
##                         (0.057)                      
##                                                      
## log_violent_crime                        0.291***    
##                                          (0.045)     
##                                                      
## log_pop                 0.174***         0.531***    
##                         (0.060)          (0.054)     
##                                                      
## Constant               -1.400***        -1.610***    
##                         (0.146)          (0.185)     
##                                                      
## -----------------------------------------------------
## Observations              216              214       
## R2                       0.828            0.755      
## Adjusted R2              0.826            0.753      
## Residual Std. Error 0.209 (df = 213) 0.246 (df = 211)
## =====================================================
## Note:                     *p<0.1; **p<0.05; ***p<0.01

Fact #10 Liberal Places do not have fewer cops

## 
## ===============================================
##                         Dependent variable:    
##                     ---------------------------
##                            log_officers        
## -----------------------------------------------
## demshare_pres_2016             0.135           
##                               (0.085)          
##                                                
## log_violent_crime             0.099**          
##                               (0.039)          
##                                                
## log_pop                      0.875***          
##                               (0.052)          
##                                                
## Constant                     -2.900***         
##                               (0.183)          
##                                                
## -----------------------------------------------
## Observations                    205            
## R2                             0.849           
## Adjusted R2                    0.846           
## Residual Std. Error      0.191 (df = 201)      
## ===============================================
## Note:               *p<0.1; **p<0.05; ***p<0.01