Since its inception, NARPAC has resisted using state and county data to provide benchmarks to determine the efficacy of DC's strictly urban budget. However, comparable data for other cities is relatively hard to come by, and NARPAC has frequently resorted to comparing DC's government costs with those of its surrounding suburbs. Census 2000 data has been used to make overall comparisons and comparisons between kids in the various jurisdictions of this metro area. But the suspicions that the DC government is relatively inefficient have not been dispelled by any of the earlier comparisons. New analyses have now been made using Census 2000 data in an effort to determine which parameters do or do not provide useful trending information:
o The distribution of state and local spending is quite consistent among the 50 states, although there is a significant difference from the highest to lowest spending in terms of percent of total personal income;
o The level of government employees per capita shows that DC is very different than its neighboring states or the US as whole;
o Public Education is one of the major elements in all state/local spending, and NARPAC has made new scatter plots to determine trends among 40 large urban school districts with a median size very close to DC's. The District is nowhere near the norm;
o Public Safety is another major cost factor, and NARPAC has sorted through equivalent data for 40 major American cities. DC is outside the norm for police per capita, though the incidence of serious crime is not;
o An important role of minority shares in urban government costs has always been assumed, but cannot be confirmed across the 50 states;
o One reason for avoiding the use of state-level data has been the assumed differences in rural and urban demands on state/local government spending. A plot of major cost elements as a function of each state's urbanization dispels that assumption, and further suggests that there is no consistent difference in the unit costs of providing major government services;
o On this basis, DC's departure from the norm appears very substantial;
o Additional analysis of twenty US cities was conducted as part of NARPAC's refutation of a recent GAO report exaggerating DC's 'structural imbalance' which again show DC outside the norm;
o A thumbnail sketch of five diverse cities in the news in the Fall of 2003 again shows DC's departures from the urban norms: Richmond, VA and Oakland, CA are cities previously managed by DC's new City Administrator, and Charlotte, NC and El Paso TX, have just moved ahead of DC in city population.
o In a similar vein, DC is compared statistically with Long Beach, CA, in anticipation (wrongly, it turns out) of the possible hiring of that city's recent school superintendent. Based on demographic differences, it is doubtful that Carl Cohn could have repeated his success here.
o There is a tremendous impact from teenage births on national and local poverty, poor school performance and poor health, and crime. There are encouraging trends in reduced birth rates nationally and in the US, and new pharmaceutical developments may generate important ways to break the cycle of poverty.
This first chart displays the major components in state and local government spending for each of the 50 states, the US as a whole, and the District of Columbia. The elements of spending are lumped together to indicate the major categories. Public education includes both secondary education and higher education. Welfare includes all components of human services, including housing and community development. Safety includes the usual aggregation of police, fire and emergency, judicial and corrections. Overhead includes the administrative elements and interest on debt. "The rest"category is a catch-all for public works, economic development, etc. This analysis explores the first three as "independent variables", assuming the latter two can be treated as overlaid percentages (averaging 12% and 23% respectively).
The three highlighted bars show DC near the top, the US total in the middle, and the simple
average of each component at the bottom. What is noteworthy of course is the departure of the
DC bar from all the rest. Not only is it near the top of stack in terms of percent of total personal
income for the jurisdiction (22%), but the proportions of its major subcategories are anomalous.
Why are welfare and safety so large, and education relatively so low? Subsequent sections
to explain this.
GOVERNMENT EMPLOYEES PER CAPITA
This next chart approaches the same problem from a different vantage point: the number of local and state government employees per capita for a far more fine-grained set of spending components, using DC, its two neighboring states, and the US (a foursome used in many other comparisons throughout this web site). What is striking here is how different many of the components are. Again, while the two states quite closely match the aggregate total of all the states of the US, DC stands apart for several reasons that impact on further analysis.
While central administration and financial services require somewhat
more personnel than in the states and US, the "safety" set (all shades
of blue) is almost double in size, and the "welfare" set (autumn colors?)
is more than twice as large. The "other" category (striped) is perhaps
proportionately about the same, but the "education" set is significantly
different. While the secondary education lines (teachers and support)
are a bit smaller than for the states, the higher education slices are
almost non-existent, reflecting DC's lack of a major public college/university.
Hence DC employs some 90 employees per 10,000 residents more than the
states, there are about 50 less in education. It is also clear that DC
employs an unusual number of support personnel compared to their teaching
The four scatter charts above may not be precise enough to satisfy mathematical purists, but each shows some interesting trends. The scatter points are "boxed in" to a generalized trend line for each set of compared parameters. The first point to notice is the size of the scatter. The second point is that despite this scatter, trend lines are quite evident. Third, blue symbols are used for school districts less than 50% minority students, and green for the rest. DC is awarded a red circle to make its location inescapable. Most significant in this third area is that the student mix has very little to do with the overall trend lines.
In the upper left chart, there is an unmistakable trend across 40 large school districts that school systems with more, smaller schools (less students per school) cost more per student than larger schools. This fits NARPAC's instincts for 'economy of scale' as postulated in a typical school personnel pyramid. Note that DC is roughly "in the middle" of per student costs for the smallest schools, but might reasonably save $2000 per student by going to fewer, larger schools. This might also improve the poor teacher-to- support ratio noted in the prior section.
The lower left chart demonstrates that DC tends to have the worst student-to-teacher ratio of these 40 large urban districts, in large measure because its schools are too small.
On the other hand, the upper right hand chart is important because there is no trend line between drop-out rates (or conversely graduating fraction of school body) and school size. DC is a little below par in graduation rates, but this chart would say it would not be effected by school size.
Finally, the problem of drop-our rate vs racial composition of the school system appears to be a fact of life. DC is "well within the scatter" for a largely minority student population. As NARPAC concludes elsewhere, this lower performance is not so much to do with race and poverty as it is with parental presence and their education levels. To make this point again, the charts below take the 31 Maryland and Virginia counties with school systems larger than 10,000 students, and plots the relationship between household income and both education level and poverty level. Clearly poverty and median household income are related, but more directly, household income drops linearly with decreasing "education factor". This "education factor" is simply the percentage with high school degrees, plus the percentage with college degrees of any sort (i.e., no one in the county with a high school degree = 0, everyone with a college degree = 200).
These next three charts show basic trends among forty major American cities (excluding the very largest like New York, Los Angeles and Chicago, which might show anomalies). The X axis across the bottom of all three shows the minority percentage in the population of each city, running from 20% to85%, since this is generally accepted as an important variable in public safety.
The top chart looks at serious crimes per capita (not just violent crimes) and can only conclude that they are not related to urban minority populations!. And in fact, DC seems to be just about at the norm, despite its high fraction of minorities.
The middle chart graphs the number of uniformed police per capita in each city, and finds a significant trend. Cities with 80% minorities are likely to have 2.5x as many cops in uniform as cities with 20% minorities. But cause and effect cannot be discerned from this: does it take 2.5x as many police to keep the crime rate down, or do largely minority cities just pad their police forces? In either event, DC is off the scale here. It would fall in the center of the scatter chart with 40% fewer police in uniform.
The bottom chart repeats the same information but includes the civilians in the police
DC's share of civilians on the force is not abnormal, but this graph would suggest that there are
probably 40% too many civilians on the force as well. This is particularly true since DC has
police in other, mostly federal, agencies than in its own force (incl., Park Service, Secret Service,
Capitol Police, GSA police, Metro police, and so forth).
These four charts all relate to the supposed influence of the minority share of a jurisdiction's population, in this case the 50 US states which run from two or three percent minority to about 45%. It is in this parameter that DC seems in a class by itself relative to states, though not to cities or urban school districts.
The upper chart shows the unmistakable relationship between minorities and poverty, and suggests that, if anything, DC is on the low side of the poverty contingent, given its high percentage of minorities. But what is more important, in the next three categories there is no correlation between minorities and the three most important determinants of state and local government spending:
The cost per kid (second chart) in secondary education has considerable scatter, but no trend line. But again, DC appears a good 20% high (explained earlier as the cost of small schools).
In the third chart, the dollars per welfare recipient (all categories) shows a large variation, but it surely is not related to minority percentages. But on this chart, DC's welfare spending appears to be more than twice as high as the average!
On the bottom chart, the total per capita state/local costs of providing safety are plotted for those
populations living in metro areas, again as a function of minority fraction. And again there is no
trend line: minorities do not influence safety costs (!), and DC's costs appear more than twice as
high as they should be.
It is also conventional wisdom that urban concentrations of residents are likely to incur higher state/local government costs than their rural counterparts. This chart tests this hypothesis for the fifty states which run from under 30% living in metro areas to New Jersey where everyone lives in a metro area. It plots the sum of the per kid school costs, per-recipient welfare costs, and per metro area resident safety costs. Clearly, there is no correlation whatsoever with urbanization, but equally clearly, DC costs are out of line from the average by at least 40%. NARPAC concludes from this that even though DC is surely "a city, not a state", state- aggregated data can be used to suggest those areas where DC government costs are out of line with the American norm.
In the Fall of 2003, four cities, seldom compared directly to DC came into the local news for two very different reasons:
DC's brand new City Administrator, Robert C. Bobb, had previously served as the City Manager of Richmond, VA for eleven years, followed by a similar position in Oakland, CA for six years. NARPAC wondered how these two very different cities compare to the nation's capital city, and whether they prepare Mr. Bobb for his challenging new task.
According to the Census update of 2002, two more American cities now exceed DC in population, reducing the District to 23rd place, even those it remains the core city of the nation's fourth most populous metro area. These two very different, but rapidly growing, cities are Charlotte, NC and El Paso, TX. NARPAC also wondered what characteristics might be contributing to their relative growth.
The chart below (click-up for legibility) presents 20 different comparative parameters, "normalized" to the base value for DC so they can be shown on the same charts:
In (public) school demographics, minorities comprise over 50% of the enrollment, and generally exceed the equivalent population distribution of their parents. Every city has more kids- per-school, with the faster growing cities having schools 75% larger or more.. Kids per teacher also is higher for all cities except Richmond, and kids per non-teaching staff member are also higher. DC schools again appear top-heavy with personnel costs.
Under household demographics, all cities have more family households, and all cities save Richmond, have far fewer families headed by a single parent. DC overall "AgEd" education scores (a combined score for percent of population with high school degrees or better and college degrees or better -- please see chapter on education, poverty, and ignorance) are equal or better than the other cities. As demonstrated elsewhere on this web site, median household income is closely related to that parental education level.
Regarding crime statistics, total reported crimes of all kinds are quite closely related per capita. Violent crimes are significantly lower only in El Paso, but the murder rate are very substantially lower than for all cities except Richmond. On the other hand, as has been pointed out ad nauseum all over this web site and particularly regarding DC's supposed structural imbalance, DC spends well over twice as much on police protection as any of the other cities.
Finally, under government statistics DC's expenditures per capita are 50% to 80% higher than in any of those other four cities, as is the total state and local employment (except for Richmond). This is despite the fact that poverty rates in Richmond and El Paso are significantly higher than in DC. And in the only statistic not normalized to DC as a baseline, all these core cities' populations are a larger share of their metro area's than in the case of DC.
Not much significant wisdom is produced by these comparisons, although the following general conclusions would appear to be reasonable:
o Bobb's experience in Richmond would appear more closely related to DC's situation than that of Oakland;
o The faster growing cities share lower population densities and fewer minorities, more families and more two-parent families, lower murder rates (but similar overall crime rates), substantially lower government costs, and less overwhelming suburbs. Surely, however, there are other more important factors not captured by these statistics, from employment opportunities to weather.
o On the whole however, there is nothing in these statistics to indicate that in any way DC is an exemplary American core city. In fact it appears to be below the norm in most categories.
For about two months in the spring of 2004, DC had its fingers crossed that a well-known public school superintendent from Long Beach, California could be enticed to bring his California-born and -raised family to the nation's capital. By the first days of summer, and after considerable negotiation over DC's clumsy school management problems, retired Superintendent Carl Cohn decided to pass up the opportunity (at almost any salary) and to remain in his current consulting and academic position at the University of Southern California.
NARPAC, eager to be ahead of the curve with pithy analysis, is thus left with a somewhat irrelevant comparison between two cities, one quintessentially "West Coast", the other typically mid-Atlantic. It does, however, suggest that there are major differences in the culture and demography of the two cities, and that what had been sauce for the goose might not have become sauce for the gander.
Long Beach, though a separately incorporated city, is a relatively indistinguishable part of the Los Angeles metro area. Unlike DC, it is not the "core city", or for that matter, the major calling card, for its area. It is bordered on the East by Los Angeles, on the North by the tonier Huntington Beach, and on the West by the Pacific Ocean. It provides the major port for the area Long Beach, with some The tables below compare the two cities by converting current statistics on DC to match the relevant data for Long Beach. In fact, Long Beach is considerably smaller than DC, with a population of 462,000 ( in '00) and an area of under 51 sqmi.. Whereas DC ranked 21st in US city population, Long beach ranked 34th in the 2000 Census. The bigger difference, however, is that Long Beach is growing in population while DC has yet to reach its nadir. In fact, by 2003, DC had dropped to 25th, and Long Beach had risen to 32nd. Here are some of the relevant comparisons from the first chart:
If DC, at its current population level, had Long Beach's gender mix, it would 11,500 more males and the same number fewer females, a 4% shift which might have only minor impact on the relevant sociological mix. On the other hand, it would have almost 52,000 (45%) more kids, balanced by almost 34,000 fewer adults, and an important 26% (18,000) fewer seniors over 65, who are typically large consumers of city services. Long Beach's racial demographics would replace over 258,000 blacks with almost 160,000 Hispanics, 54,000 Asians, and 30,000-odd whites.
Household composition (also left column below) would be significantly different in DC as well. Because Latino families tend to be larger, DC, at its current population level, would have almost 42,000 fewer households. However, DC would have some 12,000 more family units, 23,000 more with kids and almost 24,000 more married families with kids, an increase of well over 100%, and 3000 fewer single moms. From NARPAC's point of view, these last two statistics are likely to result in stronger performance in school, despite a significantly increased number of DC families (5,300) and families with kids (6,600) in poverty.
school enrollment and costs
With respect to public schools (right column above), instead of a substantial drop in school enrollment since 1990, DC schools would have 37,000 (54%) more kids. 47,000 fewer kids would be black, 25,000 more would be Hispanic, and almost 20,000 more whites and Asians. Far more important from the standpoint of both school costs and, to a lesser extent, classroom decorum, there would be over 70% (almost 3000) fewer "ungraded" special ed kids, and DC's graduating class in 2000 would have been 1,300 larger.
Equally interesting, in juxtaposing Long Beach, a city with 30 years of rising school-age enrollment, on DC, with its continuously declining youth population, DC would have 76 fewer school buildings, including 68 less primary schools and nine less high schools. At its current school enrollment, however, it would be spending 51% less ($441M) in tot, but only 18% less ($56M) on classroom instruction. These lower costs, for whatever reason (surely including fewer facilities and many fewer special ed kids), would reflect in a $4,800 (44%!) lower spending per kid, including $481 (32%) lower Title I grants per poor kid. These are very substantial cost differences indeed, but they are not necessarily due primarily to magic management moxie.
parental education and labor force mix
Meanwhile, (next chart below, left column), the parental educational demographics would be quite significantly different. Most important, DC, with its current population level and household size, would have almost 10,000 (46%) more fathers involved in the education of their kids, and a few less moms. There would be 25,000 (84%) more adults (not necessarily parents) with less than a 9th grade education, presumably mostly recent Latino immigrants, but almost 12,000 more attending or completing high school. There would be 45,000 (65%) more with some college education, but this is only because there would be some 58,000 less students that received or went beyond a college bachelor's degree.
As also shown below, there would be a significantly different labor force mix, with 26,400 (39%) fewer workers with the relative security of government jobs, offset by 22,600 more relatively competitive (and economy-sensitive) private sector jobs, plus 3,400 more self-employed entrepreneurs. This might produce some significant differences in overall work ethic and a greater emphasis on governmental efficiency, but NARPAC has no analytical background for elaborating on this. Similarly, the occupational mix tends further towards some 44,000 more commercial service industry jobs and a much larger share of hands-on workers in construction, production, and transportation.
There are other substantial differences in urban statistics that may also set a somewhat different urban environment. (right column above). If DC's population had Long Beach's crime rates, there would be very substantial differences: 66% (159) fewer homicides; 53% (4,600) fewer violent crimes; and 44% (14,600) fewer property crimes: for a grand total of 46% (19,500) fewer "serious" crimes. And this would reflect in a 71% reduction in uniformed police officers. These crime reductions are even more significant when it is noted that these lower crime rates reflect a city with many more teenagers and considerably more families in poverty. Other forces must be at work in the neighborhoods.
the great American love affair
Of perhaps less interest sociologically, but of more interest environmentally, If DC's present population had Long Beach's attachment to automobiles, almost 91,000 (70%) more workers would be commuting by car, and over 66,000 (77%) fewer would use public transit. In terms of ownership, almost 53,000 (58%) fewer households would be without cars, while 38,000 more )twice as many) would have two cars, and three times as many (18,600) households would have three or more cars. In toto, DC would have almost twice as many (103,500) more cars registered within its city limits. (Vrooom!)
economic comparisons There are some conflicting comparisons in household income and owned-home value between the two cities, as shown on the charts to the left (Long Beach, red bars, DC, green bars). DC has more very poor (under $10K) and very rich (all over $100K) residents, and somewhat fewer in all the income brackets in between (upper chart). Broadly speaking, it suggests a diminished "middle class" in DC, ranging all the way from $10K thru $75K, because of its "oversized" extremes.
In terms of owned-home value, the middle value for Long Beach ($200-$300K) is twice as high as in DC ($100-150K). This is a far larger discrepancy than would be indicated by the income distribution itself. Is this because there are significantly fewer two-adult families in DC, or are other forces at work? NARPAC cannot say, but overall the percentage of renters is almost identical between the two.
comparative school scores
Finally, NARPAC has taken a somewhat more detailed look at the relative school scores between the two continent-apart urban school systems, and the chart below may appear somewhat complicated. The standard four testing categories are displayed: reading and math at both 4th and 8th Grade levels. The five bars in each cluster provide the same comparisons, hopefully defined by their different colors. Each one shows the standard four performance subdivisions ranging from "below basic" (depicted as negative numbers), plus "basic", proficient and a critical few "advanced". Each bar sums to 100%. The balance between positive and negative is essentially the difference between winning and losing the next generation. The proportions among the winners' capabilities determines their likely lifetime ranking in the competitive American economy, and their ability to avoid slipping into, or repeating the cycle of family poverty.
The center "baseline" bar is the American average for all urban school districts, regularly significantly lower than the norm for non-urban school districts. Immediately above and below the center bar are NARPAC's calculations of the urban average adjusted for racial composition. The rationale for this is more thoroughly addressed under NARPAC's extensive section on "exploring NAEP test scores" . It provides more reasonable expectations of performance within each of the four major racial cohorts. It also avoids direct comparisons between, say, school classes that are predominantly black (or Hispanic) and those predominantly Asian (or white) which, throughout the US, are at different ends of the spectrum. In all four test cases, an average baseline for DC's particular mix is below that of the national urban norm, while Long Beach (with its higher Asian population) is a bit above the chosen baseline.
To NARPAC, the best indicator of an individual school system's performance is whether its scores surpass or lag behind the national "average" for its (NARPAC-computed) racial mix. The answer seems relatively straightforward. The Long Beach school surpasses its adjusted baseline in 4th Grade reading, but loses most of that advantage by the 8th Grade. It starts a bit behind in 4th Grade math, and slips further behind by 8th Grade. By comparison, the DC school system starts noticeably behind its realistic norm, but closes the gap somewhat among the older kids. Compared to the DC School System, somebody is apparently doing something somewhat better in the Long Beach system, particularly at the lower grades. But the difference is nowhere near as stark as indicated by directly comparing the top and bottom bars in each set.
NARPAC believes it would be dangerous to assume that the much larger gaps directly between Long Beach and DC are a realistic testimonial to West Coast school management. The "losing" kids in Long Beach schools (clearly the most basic indicator), vary from perhaps 32% to 48% of their classes. The "losing" kids in the DC system vary from 50% to 70%., and this is clearly reflected in their higher drop-out rates, crimes rates, etc. In all cases, DC lags Long Beach in "winners" by one-third to one-half. But many non-school factors are at work in these apparent differences.
There is no way that DC's near-desperate leadership should expect any ten-year Long Beach school superintendent to close those gaps in winners and losers even if he had stayed here for 10 years. On the other hand, he might have been able to eliminate some of the other internal deficiencies that diminish the quality of DC school life.
The more NARPAC looks at the multiple socioeconomic problems that beset the nation's capital city, the more crucial becomes the role of single, teen-aged mothers whose kids are odds-on more likely to be less healthy, poor learners, school drop-outs, and in trouble with the law. These mostly accidental moms are not only destined for a life of poverty themselves, but to perpetuate the cycle through their kids. Changing the culture of these teenagers is clearly the crux of rectifying the unaffordable, and globally embarrassing, societal imbalance that afflicts DC and other American inner cities.
For a variety of good reasons, teen birth rates have been declining for more than a decade. They could decline even more steeply if the FDA follows the advice of its advisory committees and, absent some ill-advised political intervention, approves making levonorgestrel as available as condoms and cough syrup. This "3-mornings-after" contraceptive pill known as "Plan B" is already available in dozens of other countries. It has the strong ideological advantage of being able to prevent conception, but unable to induce an abortion. A green light for over-the- counter (and from-the-school-nurse) availability of this drug could be a very major milestone in the gradual alleviation of American urban poverty, and all its consequences.
Even without Plan B, however, significant reductions (about 25%) in teen-age births to kids between 15 and 19 years of age have occurred between 1991 and 2002. These reductions in (mostly unwanted) babies are evident in all 50 US states, DC, and the four US territories, as shown on the chart below. It is part of DC's globally-embarrassing statistics to have the highest rates in both 1991 and 2002 (followed closely by Guam and Mississippi). Granted DC is an inner city, not a state, but there seems to be little inherent reason why DC birth rates should remain 61% above the national average, and virtually twice as high as for neighboring Maryland and Virginia.
As shown on the next two charts, black teenagers remain more than twice as likely as white teenagers to give birth (upper chart), no doubt because more of them are trapped in the cycle of urban poverty and have far more difficult parental circumstances. The lower chart shows the break-out among black moms in various age categories. What could be more encouraging than having the birth rate drop almost 50% for 15-17 year-olds, over 30% for 18-19 yr-olds, and only 20% for 20-24 yr-olds?
But despite these improvements, black moms are still on average several years younger than their white counterparts (upper chart, below), and frequently less well schooled as many fewer get the chance for a college education (lower chart):
Of some interest are fresh data (to NARPAC) on the relative ages of the teenage dads who help limit the lifelong potential of their often casual partners. The upper chart below indicates that for white procreation, the dads are only half as likely to be in the 15-19 age group themselves and are much more likely to be 4-5 years older. A virtually identical pattern is shown in the lower chart for black partners. Unfortunately, trying to keep younger women away from older and more mature young men is probably a losing game in any of the world's diverse cultures!
There is also a distinctive trends towards young black moms having more kids than their white counterparts. As shown below, black moms are about 75% more likely to have three kids before their 20th birthday, and over 30% more likely to be raising four kids before their 25th birthday:
And far more depressing to human beings and budget analysts alike, early
prenatal care is more common nationally among white moms than black, and
DC exacerbates that difference (left). As a consequence, black moms are
noticeably more likely than whites to be confronted with offspring with
congenital defects (right). More tragic fodder for the cycle of poverty
as well as municipal budgets:
Did you find this of interest?
This page was updated on July 5, 2004
© copyright 2007 NARPAC, Inc. All rights reserved