The Inland Empire has been among the economically hardest hit regions in the U.S. during the Great Recession, with unemployment rates hovering around 14 percent since 2010. The misery, however, is not evenly spread among the various cities within the region. An analysis of the thirty-six cities with population above 25,000 shows that there is substantial variation in terms of labor market performance. According to the most recently available monthly employment report (July 2011), the unemployment rate for the Inland Empire region as a whole rose to 14.7 percent. Yet six of the thirty-six cities (Murrietta, Upland, Rancho Cucamonga, Palm Desert, La Quinta, Chino Hills) have unemployment rates of 10 percent or less, while at the other extreme, there are four cities (Adelanto, San Jacinto, Perris, Coachella) with unemployment rates higher than 20 percent.
What causes such a disparity in labor market performance among the Inland Empire cities? More generally, what are the determinants of unemployment rates in this particular region? Finally, and perhaps most importantly, what policies can be used to lower the unemployment rates of cities within Inland Empire, and in doing so, what can we learn about lowering the unemployment rate in the region as a whole?
Does Size Matter?
Some have suggested that larger or more populous labor markets produce, on average, lower unemployment rates, since it is easier for job seekers and employers to find each other in a bigger city rather than in a relatively smaller city. The implication is that unemployment rates might benefit from economies of scale, and this theory has been proven empirically between MSAs in the United States. We investigate this possibility in the Inland Empire in Table 1, which ranks most recently available monthly unemployment rates of the major Inland Empire cities in descending order. The last column lists the corresponding population size.
To investigate the possibility of economies of scale, we highlighted in red cities with more than 90,000 inhabitants. Some of the larger cities, such as Temecula and Corona, fit the hypothesis, yet others (Moreno Valley, San Bernardino) clearly do not. Viewing the table as a whole, larger cities are not concentrated in the low unemployment rate section, and smaller cities are not predominantly in the high unemployment rate section. Smaller cities such as Adelanto are as likely to experience high unemployment rates as larger cities such as Victorville. In general, there is no apparent pattern visible for the highlighted cities, and this suggests that size does not matter as a determinant of unemployment rates. This size effect does not seem to be present within the Inland Empire, perhaps because these cities do not have exclusive labor markets. In fact, the high percentage of people commuting to or working outside of the city in which they reside suggests that the Inland Empire cities might share the same labor market. Therefore, although the size pattern may be observable between MSAs according to some studies, it is less likely to be found within an MSA, and it is certainly not present in the data displayed in Table 1.
The conclusion regarding the size effect within the Inland Empire is not dependent on any specific month during which we observe the labor market. In addition, it holds irrespectively whether we use monthly data or annual data.
Finding that size does not matter in the list of potential determinants of unemployment rate differences has not gotten us any closer to an explanation of the observed variation between cities. Perhaps a geographical map of the cities might lead to further clues. Figure 1 illustrates the same thirty-six cities, using darker areas to indicate higher unemployment rates. Setting aside the cities of the Coachella Valley, there appears to be a divide between cities that lie closer and those that lie further away in terms of driving distance to their respective closest “point of entry” into Los Angeles County, Orange County, and San Diego County. By “point of entry” we mean the highway exit on the county line one takes to enter the county of destination. Cities that are closer to these points have lower unemployment rates, on average. In other words, geography does seem to matter.
There currently is an East-West divide in California in terms of unemployment rates where coastal areas, such as Los Angeles, San Francisco, and San Diego, are less affected by the downturn and the slow recovery than the areas that lie further inland. Now assume that residents of communities that live closer to “points of entry” into the economically less depressed areas are more likely to commute from the Inland Empire into these areas, and hence are more likely to hold jobs there. Since unemployment rates are measured by residency (if you lose your job in downtown L.A. and reside in Ontario, the unemployment rate of Ontario goes up while the unemployment rate of L.A. is unaffected), then these communities will show lower unemployment rates when compared to those further away from the “points of entry.”
One third of those who live in the Inland Empire and hold jobs commute to Los Angeles, Orange, and San Diego counties to work. Many residents from those counties who moved to the Inland Empire were drawn by more affordable housing further inland, rather than by the lure of jobs. The combination of more affordable housing and the lack of relatively better paying local jobs in San Bernardino and Riverside counties resulted in these residents spending substantial time commuting to work in neighboring regions.
To determine whether geography matters in explaining unemployment rate differences, we display 2010 annual unemployment rate data in Table 2, with an additional column listing the distance from each city to its respective nearest “point of entry” into either Greater L.A. or San Diego County, whichever is closer. We will ignore the cities of the Coachella Valley in our analysis (marked in blue), because they are clearly too far away for regular commuting. Thus, the Coachella Valley economy must be viewed separately from the other cities when it comes to finding the determinants of unemployment rate.
After excluding cities of the Coachella Valley, we marked in red those that were within 15 miles of their “points of entry” in Table 2. As one can easily observe, the majority of the cities with shorter distance to the “points of entry” are found in the bottom part of the list. This pattern suggests a positive relationship between distances from the “points of entry” and city unemployment rates: on average, cities that are located further away will experience higher unemployment rates.
To further investigate this observation, Figure 2 presents a cross-plot of city unemployment rates against the distance from each city center to its “point of entry.” Excluding the cities of the Coachella Valley for the reasons stated above, we constructed a trend line based on data from the remaining cities.
The trend line suggests that geography does matter, but the effect becomes weaker as one moves further away from the county line. In other words, unemployment rates change more drastically in the 0 to 20 mile range from the “point of entry” than in the 20 to 40 mile range. Furthermore, this effect is economically important. Moving away from the county line for the first 20 miles, the unemployment rate increases, on average, by a massive five percentage points.
However, not all of the city unemployment rate observations lie on the trend line. In fact, the considerable scatter around the trend line indicates that there must be determinants other than proximity which play a significant role. That is, closeness to employment centers in the western counties matters, but there are limits to its explanatory power. Take Ontario and Upland as an example. While both cities are roughly the same distance away from the nearest “point of entry,” about 5 miles, Ontario’s unemployment rate, at 15.1 percent, is more than 5 percentage points higher than Upland’s 9.9 percent. Similarly, San Jacinto and Redlands are both located around 32 miles away from the county line, yet San Jacinto’s unemployment rate (21.8 percent) is more than 10 percentage points higher than Redlands’s (10.5 percent). What causes the variation in the unemployment rates given that they have the same proximity to more vibrant economic areas?
We considered a variety of city attributes: median household income level, number of housing permits issued, average household size, average education level, crime rates, demographics, and residential status (rent/own). After controlling for the influence of geography, three of these variables stand out: median per capita income level, percentage of residents with a high school diploma, and crime rates.
Table 3 compares the two city pairs mentioned above by listing the values for the three new variables. Recall from the previous text that Upland and Redlands have lower unemployment rates than Ontario and San Jacinto respectively, while being very similar in terms of proximity to “points of entry.”
Table 3 demonstrates that after controlling for geography, a higher median household income, a higher education level, and lower crime rates result in lower city unemployment rates. More complicated statistical techniques allow us to establish the separate effect of each attribute while controlling (“holding constant”) the others. Performing this type of (multiple regression) analysis establishes the following: apart from geographic factors, the unemployment rate of a city is higher, on average, if
• households have lower income;
• the percentage of high school graduation is lower.
Remarkably, higher crime rates only have a positive effect on unemployment rates when not controlling for income and education level. That is, crime rates lose their significant explanatory power when taking income and education level in addition to geography into account. Hence, crime rates do not play a separate role in determining city unemployment rates above and beyond the influence established by the other factors. To emphasize the result, once geography, income, and education are allowed to cast their effect on the unemployment rate, crime rates have no additional contribution.
What are the Policy Implications?
Given our results regarding the determinants of city unemployment rate variation, what can policy makers do to improve city and county unemployment rates? Clearly cities cannot be relocated easily, so the “closeness” geographical effect must be taken as given. This statement is less obvious than it appears at first. Greater Los Angeles, for instance, has expanded outwards dramatically over time. In other words, whereas the county line always remains at the same geographical location, the employment centers can move closer to their employees over the years.
This fact leaves the other two alternatives as sole factors that can be influenced to have an impact on unemployment: household income and high school attainment levels. Government officials have the ability to raise average household incomes by attracting higher paying jobs into their area, thereby generating higher paying employment opportunities. This can be done through enterprise zones and other subsidies and tax breaks which are under governments’ control, directly or indirectly, in policy circles at the state and local level. However, there are other obstacles to overcome before higher value-adding firms move into an area. These firms are particularly interested in hiring skilled workers, which may be problematic in certain areas of the Inland Empire, given their low education level. Take Adelanto for example. Only 63 percent of its population had a high school degree in 2009, strikingly low when compared to a national average of 85 percent and the average for California, 77 percent. Moreover, the housing boom in the Inland Empire in the late 1990s was created by households with lower income immigrating, instead of by better educated, higher income-earning families, as many lower-income households were attracted to the area by affordable housing.
Hence, it is the third factor, high school attainment levels and education in general, that plays a central role in tackling the labor market problems. Clearly, higher education levels have an effect on median household incomes, but there also seems to be an additional contribution from education beyond its impact on income. Our analysis places education in the center of policy options to reduce unemployment in the Inland Empire. One possibility that has been tried in the past, and claimed by many to be unsuccessful, is to throw money at the problem. That is, to improve education outcomes by increasing expenditures per student and/or by reducing class sizes. However, setting aside the effectiveness of these programs, such a policy is clearly not an option in the current stagnant economy. Moreover, we foresee further school budget cuts in the future.
Fortunately, there are ways to raise high school attainment rates without raising expenditures, such as promoting high performance teachers through merit raises rather than determining salaries by seniority. Unfortunately, as the resistance in the LAUSD and elsewhere indicates, current government educational policies are not implemented along such lines on a large scale. In reality, cuts in educational budgets are most often executed by forcing out more recently hired, younger, more passionate, and thus potentially better performing teachers – the usual LIFO policy. Similar to firms, which acquire some less productive workers during long-lasting expansions and are unwilling to get rid of them during prosperous times in absence of much need for fiscal discipline, schools are often unwilling to deal with less productive teachers. Laying off teachers is costly or even impossible for administrators both in terms of existing tenure rules (note that, as a general principle, tenure does not prevent districts from cutting salaries), the unpleasantness of the process for school administrators, and its negative effects on the morale of the remaining teaching staff. However, school officials should view the current economic climate as an opportune time to implement dramatic changes in school policies.
The Inland Empire saw an economic expansion of over ten years before the bursting of the housing bubble. Facing some of the most powerful unions in the United States, making educated decisions such as changing employment contracts is not only painful, but unlikely to occur during times when the need for such changes are less pressing or obvious. Now that the years of plenty have been followed by the years of famine, school districts should seriously consider how to put policies into motion that will lead to increased educational levels in the local community, not only now but also when the economy bounces back in the future.
These insights into the determinants of city unemployment rates are neither surprising nor are we the first to observe them. However, we are the first to establish these empirically for the cities within the Inland Empire. The results show that these socio-economic, geographic, and demographic factors play a significant role and have a consistent impact on city level unemployment rates within the Inland Empire. High school education, in particular, commands the most attention. Current cuts in the public sector force the government to come up with more efficient ways to operate schools. This encourages us to rethink the convention and status quo in the education system. Rewarding teachers by merit rather than by seniority can allow us to retain or even improve education quality, with the limited budget we have. The time to do it is now—- if cuts are inevitable, at least we can influence the form those cuts take.
