Metric Construction

Definition & Interpretation

\[Post\text-Depreciation \: Profitability \: Margin = \frac{Totale \: Revenue- Total \: Expenses}{Total \: Revenue} \]

Post-depreciation profit is an income measure used to determine profit after incorporating non-cash expenses on a balance sheet. Post-depreciation profit is calculated because it provides a picture of an organization’s true available profits net of depreciation expenses. Non-expense items lower an organization’s reported earnings, so a post-depreciation profit would show a lower profit in comparison to profits calculated prior to depreciation expenses.

High values in this metric are generally desirable since they indicate that an organization is not losing a lot of its revenues to expenses. Values close to zero are normal, and negative numbers indicate the organization is functioning at a deficit.

Variables

Note: This data is available only for organizations that file full 990s. [Organizations with revenues <$200,000 and total assets <$500,000 have the option to not file a full 990 and file an EZ instead.]

  • Numerator: Total Revenue - Total Functional Expenses

    • On 990: Part VIII, line 12A - Part IX, line 25A
      • SOI PC EXTRACTS: totrevenue - totfuncexpns
    • On EZ:Not available

  • Denominator: Total Revenue

    • On 990: Part VIII, Line 12A -SOI PC EXTRACTS: totrevenue

    • On EZ: Part I, line 9 -SOI PC EXTRACTS: totrevnue 



# TEMPORARY VARIABLES 
postd_profits  <- ( core$totrevenue-core$totfuncexpns)
revenue <- ( core$totrevenue)

# can't divide by zero
revenue[ revenue == 0 ] <- NA

# SAVE RESULTS 
core$postd_profitmargin <-  postd_profits / revenue
                     
# summary( core$postd_profitmargin )

Standardize Scales

Check high and low values to see what makes sense.

x.05 <- quantile( core$postd_profitmargin, 0.05, na.rm=T )
x.95 <- quantile( core$postd_profitmargin, 0.95, na.rm=T )

ggplot( core, aes(x = postd_profitmargin ) ) +  
  geom_density( alpha = 0.5) + 
  xlim( x.05, x.95 ) 

core2 <- core

# proportion of values that are negative
mean( core2$postd_profitmargin < 0, na.rm=T ) 
## [1] 0.4516837
core2$postd_profitmargin[ core2$postd_profitmargin < 0 ] <- 0

# proption of values above 200% 
mean( core2$postd_profitmargin > 50, na.rm=T ) 
## [1] 0.0003660322
core2$postd_profitmargin[ core2$postd_profitmargin > 50 ] <- 50



x.05 <- quantile( core$postd_profitmargin, 0.05, na.rm=T )
x.95 <- quantile( core$postd_profitmargin, 0.95, na.rm=T )

core2 <- core

# proportion of values that are negative
# mean( core2$der < 0, na.rm=T ) 

# proption of values above 1% 
# mean( core2$der > 5, na.rm=T ) 

# WINSORIZATION AT 5th and 95th PERCENTILES

core2$postd_profitmargin[ core2$postd_profitmargin < x.05 ] <- x.05
core2$postd_profitmargin[ core2$postd_profitmargin > x.95 ] <- x.95

Metric Scope

Tax data is available for full 990 filers, so this metric does not describe any organizations with Gross receipts < $200,000 and Total assets < $500,000. Some organizations with receipts or assets below those thresholds may have filed a full 990, but these would be exceptions.

The data have been capped to those with values between 5% and 95% of the normal distribution to cut off outliers and exempt organizations with zero profitability (though negative values are allowed still).

Reference

Any cited works here…

Descriptive Statistics

Convert all monetary variables to thousands of dollars. Scale metric up (multiply by 100) for readability.


core2 %>%
  mutate( # postd_profitmargin = postd_profitmargin * 10000,
    totrevenue = totrevenue / 1000,
    totfuncexpns = totfuncexpns / 1000, 
    lndbldgsequipend = lndbldgsequipend / 1000,
    totassetsend = totassetsend / 1000,
    totliabend = totliabend / 1000,
    totnetassetend = totnetassetend / 1000 ) %>% 
  select( STATE,  NTEE1, NTMAJ12, 
          postd_profitmargin, 
          AGE, 
          totrevenue, totfuncexpns, 
          lndbldgsequipend, totassetsend, 
          totnetassetend, totliabend ) %>%

  stargazer( type = s.type, 
             digits=0, 
             summary.stat = c("min","p25","median",
                              "mean","p75","max", "sd"),
             covariate.labels = c("Post-Depreciation Profitability Margin (x100)", "Age", 
                                  "Revenue ($1k)", "Expenses($1k)", 
                                  "Buildings ($1k)", "Total Assets ($1k)",
                                  "Net Assets ($1k)", "Liabiliies ($1k)"))
Statistic Min Pctl(25) Median Mean Pctl(75) Max St. Dev.
Post-Depreciation Profitability Margin (x100) -1 -0 0 -0 0 1 0
Age 3 22 30 32 41 95 15
Revenue (1k) -5,377 259 909 4,522 3,672 408,932 14,286
Expenses(1k) 0 263 840 4,192 3,328 382,667 13,466
Buildings (1k) -4 79 824 3,504 2,868 513,509 13,210
Total Assets (1k) -7,552 778 2,446 9,262 7,477 672,021 27,039
Net Assets (1k) -178,870 156 1,094 4,553 4,079 531,068 15,470
Liabiliies (1k) -2,707 115 816 4,709 3,133 705,623 18,722

What proportion of orgs have post-depreciation profitability margins equal to zero?

prop.zero <- mean( core2$postd_profitmargin == 0, na.rm=T )

In the sample, 0 percent of the organizations have post-depreciation profitability margins equal to zero. These organizations are dropped from subsequent graphs to keep the visualizations clean. The interpretation of the graphics should be the distributions of post-depreciation profitability margins for organizations that have positive or negative values.

###
### ADD QUANTILES
###
###   function create_quantiles() defined in r-functions.R

core2$exp.q   <- create_quantiles( var=core2$totfuncexpns,   n.groups=5 )
core2$rev.q   <- create_quantiles( var=core2$totrevenue,     n.groups=5 )
core2$asset.q <- create_quantiles( var=core2$totnetassetend, n.groups=5 )
core2$age.q   <- create_quantiles( var=core2$AGE,            n.groups=5 )
core2$land.q   <- create_quantiles( var=core2$lndbldgsequipend,    n.groups=5 )

Post-Depreciation Profitability Margin Density

min.x <- min( core2$postd_profitmargin, na.rm=T )
max.x <- max( core2$postd_profitmargin, na.rm=T )

ggplot( core2, aes(x = postd_profitmargin )) +  
  geom_density( alpha = 0.5 ) + 
  xlim( min.x, max.x  ) +
  xlab( variable.label ) +
  theme( axis.title.y=element_blank(),
         axis.text.y=element_blank(), 
         axis.ticks.y=element_blank() )

Post-Depreciation Profitability Margin by NTEE Major Code

core3 <- core2 %>% filter( ! is.na(NTEE1) )
table( core3$NTEE1) %>% sort(decreasing=TRUE) %>% kable()
Var1 Freq
Housing 2837
Community Development 1585
Human Services 1102 

t <- table( factor(core3$NTEE1) ) 
df <- data.frame( x=Inf, y=Inf, 
                  N=paste0( "N=", as.character(t) ), 
                  NTEE1=names(t) )

ggplot( core3, aes( x=postd_profitmargin ) ) + 
  geom_density( alpha = 0.5) + 
  # xlim( -0.1, 1 ) +
  labs( title="Nonprofit Subsectors" ) + 
  xlab( variable.label ) + 
  facet_wrap( ~ NTEE1, nrow=1 ) + 
    theme_minimal( base_size = 15 )  + 
    theme( axis.title.y=element_blank(),
           axis.text.y=element_blank(), 
           axis.ticks.y=element_blank(),
           strip.text = element_text( face="bold") ) +  # size=20 
  geom_text( data=df, 
             aes(x, y, label=N ), 
             hjust=2, vjust=3, 
             color="gray60", size=6 )

Post-Depreciation Profitability Margin by Region

table( core2$Region) %>% kable()
Var1 Freq
Midwest 1444
Northeast 1368
South 1610
West 1088 
t <- table( factor(core2$Region) ) 
df <- data.frame( x=Inf, y=Inf, 
                  N=paste0( "N=", as.character(t) ), 
                  Region=names(t) )

core2 %>% 
  filter( ! is.na(Region) ) %>% 
  ggplot( aes(postd_profitmargin) )  + 
    geom_density( alpha = 0.5 ) + 
    xlab( "Census Regions" ) + 
    ylab( variable.label ) +
    facet_wrap( ~ Region, nrow=3 ) + 
    theme_minimal( base_size = 22 )  + 
    theme( axis.title.y=element_blank(),
           axis.text.y=element_blank(), 
           axis.ticks.y=element_blank() ) + 
    geom_text( data=df, 
             aes(x, y, label=N ), 
             hjust=2, vjust=3, 
             color="gray60", size=6 )

table( core2$Division ) %>% kable()
Var1 Freq
East North Central 1038
East South Central 289
Middle Atlantic 904
Mountain 303
New England 464
Pacific 785
South Atlantic 900
West North Central 406
West South Central 421 
t <- table( factor(core2$Division) ) 
df <- data.frame( x=Inf, y=Inf, 
                  N=paste0( "N=", as.character(t) ), 
                  Division=names(t) )

core2 %>% 
  filter( ! is.na(Division) ) %>% 
  ggplot( aes(postd_profitmargin) )  + 
    geom_density( alpha = 0.5 ) + 
    xlab( "Census Sub-Regions (10)" ) + 
    ylab( variable.label ) +
    facet_wrap( ~ Division, nrow=3 ) + 
    theme_minimal( base_size = 22 )  + 
    theme( axis.title.y=element_blank(),
           axis.text.y=element_blank(), 
           axis.ticks.y=element_blank() ) + 
    geom_text( data=df, 
             aes(x, y, label=N ), 
             hjust=2, vjust=3, 
             color="gray60", size=6 )

Post-Depreciation Profitability Margin by Nonprofit Size (Expenses)

ggplot( core2, aes(x = totfuncexpns )) +  
  geom_density( alpha = 0.5 ) + 
  xlim( quantile(core2$totfuncexpns, c(0.02,0.98), na.rm=T ) )

core2$totfuncexpns[ core2$totfuncexpns < 1 ] <- 1
# core2$totfuncexpns[ is.na(core2$totfuncexpns) ] <- 1

if( nrow(core2) > 10000 )
{
  core3 <- sample_n( core2, 10000 )
} else
{
  core3 <- core2
}

jplot( log10(core3$totfuncexpns), core3$postd_profitmargin, 
       xlab="Nonprofit Size (logged Expenses)", 
       ylab=variable.label,
       xaxt="n", xlim=c(3,10) )
axis( side=1, 
      at=c(3,4,5,6,7,8,9,10), 
      labels=c("1k","10k","100k","1m","10m","100m","1b","10b") )

core2 %>% 
  filter( ! is.na(exp.q) ) %>% 
  ggplot( aes(postd_profitmargin) )  + 
    geom_density( alpha = 0.5) + 
    labs( title="Nonprofit Size (logged expenses)" ) + 
    xlab( variable.label ) +
    facet_wrap( ~ exp.q, nrow=3 ) + 
    theme_minimal( base_size = 22 )  + 
    theme( axis.title.y=element_blank(),
           axis.text.y=element_blank(), 
           axis.ticks.y=element_blank() )

Post-Depreciation Profitability Margin by Nonprofit Size (Revenue)

ggplot( core2, aes(x = totrevenue )) +  
  geom_density( alpha = 0.5 ) + 
  xlim( quantile(core2$totrevenue, c(0.02,0.98), na.rm=T ) ) + 
  theme( axis.title.y=element_blank(),
           axis.text.y=element_blank(), 
           axis.ticks.y=element_blank() )

core2$totrevenue[ core2$totrevenue < 1 ] <- 1

if( nrow(core2) > 10000 )
{
  core3 <- sample_n( core2, 10000 )
} else
{
  core3 <- core2
}

jplot( log10(core3$totrevenue), core3$postd_profitmargin, 
       xlab="Nonprofit Size (logged Revenue)", 
       ylab=variable.label,
       xaxt="n", xlim=c(3,10) )
axis( side=1, 
      at=c(3,4,5,6,7,8,9,10), 
      labels=c("1k","10k","100k","1m","10m","100m","1b","10b") )

core2 %>% 
  filter( ! is.na(rev.q) ) %>% 
  ggplot( aes(postd_profitmargin) )  + 
    geom_density( alpha = 0.5 ) + 
    labs( title="Nonprofit Size (logged revenues)" ) + 
    xlab( variable.label ) +
    facet_wrap( ~ rev.q, nrow=3 ) + 
    theme_minimal( base_size = 22 )  + 
    theme( axis.title.y=element_blank(),
           axis.text.y=element_blank(), 
           axis.ticks.y=element_blank() )

Post-Depreciation Profitability Margin by Nonprofit Size (Net Assets)

ggplot( core2, aes(x = totnetassetend )) +  
  geom_density( alpha = 0.5) + 
  xlim( quantile(core2$totnetassetend, c(0.02,0.98), na.rm=T ) ) + 
  xlab( "Net Assets" ) +
    theme( axis.title.y=element_blank(),
           axis.text.y=element_blank(), 
           axis.ticks.y=element_blank() )

core2$totnetassetend[ core2$totnetassetend < 1 ] <- NA

if( nrow(core2) > 10000 )
{
  core3 <- sample_n( core2, 10000 )
} else
{
  core3 <- core2
}

jplot( log10(core3$totnetassetend), core3$postd_profitmargin, 
       xlab="Nonprofit Size (logged Net Assets)", 
       ylab=variable.label,
       xaxt="n", xlim=c(3,10) )
axis( side=1, 
      at=c(3,4,5,6,7,8,9,10), 
      labels=c("1k","10k","100k","1m","10m","100m","1b","10b") )

core2$totnetassetend[ core2$totnetassetend < 1 ] <- NA
core2$asset.q <- create_quantiles( var=core2$totnetassetend, n.groups=5 )

core2 %>% 
  filter( ! is.na(asset.q) ) %>% 
  ggplot( aes(postd_profitmargin) )  + 
    geom_density( alpha = 0.5 ) + 
    labs( title="Nonprofit Size (logged net assets, if assets > 0)" ) + 
    xlab( variable.label ) + 
    ylab( "" ) + 
    facet_wrap( ~ asset.q, nrow=3 ) + 
    theme_minimal( base_size = 22 )  + 
    theme( axis.title.y=element_blank(),
           axis.text.y=element_blank(), 
           axis.ticks.y=element_blank() )

Total Assets for Comparison

core2$totassetsend[ core2$totassetsend < 1 ] <- NA
core2$tot.asset.q <- create_quantiles( var=core2$totassetsend, n.groups=5 )

if( nrow(core2) > 10000 )
{
  core3 <- sample_n( core2, 10000 )
} else
{
  core3 <- core2
}

jplot( log10(core3$totassetsend), core3$postd_profitmargin, 
       xlab="Nonprofit Size (logged Total Assets)", 
       ylab=variable.label,
       xaxt="n", xlim=c(3,10) )
axis( side=1, 
      at=c(3,4,5,6,7,8,9,10), 
      labels=c("1k","10k","100k","1m","10m","100m","1b","10b") )

ggplot( core2, aes(x = totassetsend )) +  
  geom_density( alpha = 0.5) + 
  xlim( quantile(core2$totassetsend, c(0.02,0.98), na.rm=T ) ) + 
  xlab( "Net Assets" ) +
    theme( axis.title.y=element_blank(),
           axis.text.y=element_blank(), 
           axis.ticks.y=element_blank() )

core2 %>% 
  filter( ! is.na(tot.asset.q) ) %>% 
  ggplot( aes(postd_profitmargin) )  + 
    geom_density( alpha = 0.5 ) + 
    xlab( "Nonprofit Size (logged total assets, if assets > 0)" ) + 
    ylab( variable.label ) +
    facet_wrap( ~ tot.asset.q, nrow=3 ) + 
    theme_minimal( base_size = 22 )  + 
    theme( axis.title.y=element_blank(),
           axis.text.y=element_blank(), 
           axis.ticks.y=element_blank() )

Post-Depreciation Profitability Margin by Nonprofit Age

ggplot( core2, aes(x = AGE )) +  
  geom_density( alpha = 0.5 )  

core2$AGE[ core2$AGE < 1 ] <- NA

if( nrow(core2) > 10000 )
{
  core3 <- sample_n( core2, 10000 )
} else
{
  core3 <- core2
}

jplot( core3$AGE, core3$postd_profitmargin, 
       xlab="Nonprofit Age", 
       ylab=variable.label ) 

core2 %>% 
  filter( ! is.na(age.q) ) %>% 
  ggplot( aes(postd_profitmargin) )  + 
    geom_density( alpha = 0.5 ) + 
    labs( title="Nonprofit Age" ) + 
    xlab( variable.label ) +
    ylab( "" ) +
    facet_wrap( ~ age.q, nrow=3 ) + 
    theme_minimal( base_size = 22 )  + 
    theme( axis.title.y=element_blank(),
           axis.text.y=element_blank(), 
           axis.ticks.y=element_blank() )

Post-Depreciation Profitability Margin by Land and Building Value

ggplot( core2, aes(x = lndbldgsequipend )) +  
  geom_density( alpha = 0.5 )  

core2$lndbldgsequipend[ core2$lndbldgsequipend < 1 ] <- NA
if( nrow(core2) > 10000 )
{
  core3 <- sample_n( core2, 10000 )
} else
{
  core3 <- core2


jplot( log10(core3$lndbldgsequipend), core3$postd_profitmargin, 
       xlab="Land and Building Value (logged)", 
       ylab=variable.label,
       xaxt="n", xlim=c(3,10) )
axis( side=1, 
      at=c(3,4,5,6,7,8,9,10), 
      labels=c("1k","10k","100k","1m","10m","100m","1b","10b") )
}

core2 %>% 
  filter( ! is.na(land.q) ) %>% 
  ggplot( aes(postd_profitmargin) )  + 
    geom_density( alpha = 0.5 ) + 
    labs( title="Land and Building Value" ) + 
    xlab( variable.label ) +
    ylab( "" ) +
    facet_wrap( ~ land.q, nrow=3 ) + 
    theme_minimal( base_size = 22 )  + 
    theme( axis.title.y=element_blank(),
           axis.text.y=element_blank(), 
           axis.ticks.y=element_blank() )

Save Metrics

core.postd_profitmargin <- select( core, ein, tax_pd, postd_profitmargin )
saveRDS( core.postd_profitmargin, "03-data-ratios/m-09-post-d-profit-margin.rds" )
write.csv( core.postd_profitmargin, "03-data-ratios/m-09-post-d-profit-margin.csv" )









Metrics generated by Jesse Lecy and Nathan Dietz
Content licensed under Creative Commons: (CC BY-NC-SA 4.0)
Source code is available on GitHub

The work is supported by the
National Center for Charitable Statistics
at the Urban Institute