[imgcontainer] [img:oklahomadesert530.jpg] [source]Becky McCrae[/source] This is Little Sahara State Park in Woods County, OK: in Woods County broadband adoption increased dramatically between 1998 and 2008 and so did local sales tax collections. [/imgcontainer]
Many politicians and main street / chamber of commerce organizations have been leery of increased Internet usage in their areas, fearing that local customers will begin shopping online and ultimately diminish the local tax base. Using Oklahoma data to study the relationship between broadband adoption and local retail sales tax collections, we found that increased broadband does not, in fact, lead to lower tax collections.
Instead, we found that tax rates and poverty rates are the features that markedly affect whether sales tax collections rise or fall.
As broadband Internet access becomes more common across the country, this technology has generally been heralded as positive for opportunities it can offer in commerce, education, and entertainment. But the impact of broadband on local retailing– and on the tax income derived from local sales – have been unclear.
Local retail stores often compete with online vendors that can provide many of the same goods. Since most online sales do not face an effective sales tax, some community developers worry that local sales tax revenues will decline as more and more people use the Internet. Loss of tax revenue could lead to curtailing many public services, such as law enforcement or health infrastructure. In fact, rural hospitals across the nation are often dependent on city or county-level sales taxes passed specifically to help them stay open.
Only some online sales are subject to sales tax: unless a store has a presence (or “nexus”) in a particular state, it is not required to gather and remit sales tax in that state. Consequently, there are no sales taxes paid on many e-commerce transactions. In fact, it is common for some people to use the Internet in order to avoid paying sales tax. Economist Austan Goolsbee (former chairman of the council of Economic Advisors) found that people who live in areas with high taxes are much more likely to buy things online.
However, e-commerce is only a small part of traditional retail sales. The U.S. Census Bureau, which collects data on e-commerce, found that while online sales have steadily increased since 2000, e-commerce made up only a small percentage of total retail sales in 2008, just 3.6%.
Given that e-commerce represents such a small percentage of the nation’s retailing, several researchers, Goolsbee included, have estimated the loss of sales tax collections to be small. But with so much at stake for local businesses and for communities should tax revenues decline, significant concern still exists about the effects of e-commerce. Last year legislation known as the Main Street Fairness Act was introduced to the U.S. House of Representatives focusing on both the potential tax revenue lost and the issue of equity between online and bricks-and-mortar stores.
While most analysis has considered how broadband access might hurt local retailers, there is evidence to suggest that broadband has both advantages and disadvantages for local stores.
Why Broadband Might Hurt Local Retailers
In addition to the price advantage of not having to pay sales tax noted above, there are a number of other reasons why consumers might prefer online retailing to shopping at more traditional physical locations. These include the ability to shop for lower prices among many competitors, convenience, access to previously unavailable goods, time savings, or simply avoiding the hassle associated with crowds and travel.
Recent statistics back this up, with 93% of Internet users indicating that they had performed some type of activity related to e-commerce. A 2010 study found that higher levels of broadband adoption have led to higher likelihoods of purchasing online.
Why Broadband Might Help Local Retailers
Yet some surveys have suggested that local retail stores might actually benefit from higher levels of broadband adoption. Many consumers use the Internet to help them shop locally, an activity known as “research online, buy offline.” Surveys conducted by BIGresearch, Forrester, and comScore have all found a high incidence of this manner of shopping. Consumers may browse and buy this way for any number of reasons, including the desire to see or touch an item before purchasing, immediate gratification, or support of local establishments.
How Do You Figure Out What Role Broadband Plays?
In order to determine whether increasing levels of broadband have impacted sales tax collections, we looked at data from two distinct periods. In 1998, broadband access was a rarity – in fact, less than 5 percent of households had access when the first surveys on the topic were conducted in 2000 (Pew Internet, 2010). Therefore, we assumed that broadband use was negligible in 1998. A decade later, however, broadband access had become widespread, with over 50 percent of households adopting high-speed Internet technology by 2008. Moreover, data on county-level broadband adoption became available for 2008 from the Federal Communications Commission. FCC’s data is broken into five categories that represent the number of fixed broadband connections per 1,000 households, lowest to highest:
• Category 1: 0 < x <= 200 (0 – 20%)
• Category 2: 200 < x <= 400 (20 – 40%)
• Category 3: 400 < x <= 600 (40 – 60%)
• Category 4: 600 < x <= 800 (60 – 80%)
• Category 5: 800 < x (80 – 100%)
Figure 1 shows the level of broadband adoption in Oklahoma in 2008. Some areas, notably those around Oklahoma City, show very high levels of adoption, while others (particularly in the southeastern part of the state) have very low levels.
[imgcontainer] [img:broadband-adoption-in-OK530.jpg] [source]Whitaker/Brooks[/source] Broadband adoption in Oklahoma, as in much of the nation, jumped between 1998 and 2008. The above map shows adoption rates by county in 2008; darker shades show higher rates of broadband Internet use. [/imgcontainer]
Figure 1. Residential Broadband Connections per 1,000 Households in Oklahoma, 2008
Since broadband access was virtually nonexistent in 1998, Figure 1 indicates that certain parts of the state adopted broadband at higher rates than others. If broadband access did have some type of relationship with the amount of retail sales tax collections, it should become apparent by observing the changes in retail sales tax collections over this period: areas with high rates of adoption should show declines in retail sales taxes, or at least lower increases relative to areas with low rates of broadband adoption.
The next two figures show per-capita retail sales tax collections in Oklahoma’s 77 counties in these two time periods: Figure 2a shows 1998 revenues and Figure 2b the 2008 collections. (the 1998 values have been adjusted for inflation to 2008 levels).
[imgcontainer right] [img:OK-County-Sales1998530.jpg]
[source]Whitaker/Brooks[/source] Figure 2a: County sales tax
revenues in Oklahoma, 1998 [/imgcontainer]
Even after adjusting for inflation, the state average increased
significantly over this time (from $186 per capita to $226 per capita).
However, most counties with high sales tax collections in 1998 still
had high levels in 2008. Similarly, counties with low collections in
1998 typically repeated that pattern in 2008.
[imgcontainer right] [img:OK-salestax2008530.jpg] [source]Whitaker/Brooks[/source] Figure 2b: County sales tax revenues in Oklahoma, 2008 [/imgcontainer]
Figure 3 takes this analysis a step further and shows the change in per capita retail sales tax collections between 1998 and 2008. Some counties declined in the amount of per capita collections (shown in white), although most increased (shown in gray) and several had significant levels of growth (shown in black). Ultimately, comparing the changes in sales tax collections in Figure 3 to changes in rates of broadband adoption in Figure 1 will help answer whether broadband access has an effect on retail sales tax income.
[imgcontainer] [img:change-in-ok-sales-tax530.jpg] [source]Whitaker/Brooks[/source] This map shows the change in per capital sales tax collections between 1998 and 2008. Darker shades show higher per capita revenues from sales tax. [/imgcontainer]
Figure 3. Change in Per-capita Retail Sales Tax Collections, 1998 – 2008 (constant 2008 dollars)
Figures 1 and 3 provide little evidence that there is any significant relationship between broadband adoption rates and changes in retail sales tax collections. Some counties with high broadband rates experienced growth in tax collections, while others saw declines. Similarly, counties with low rates of broadband adoption showed both improvements and drops in tax collections, without any apparent pattern.
[imgcontainer left] [img:Dewey-Western-Heritage340.jpg] [source]The Wildwood[/source] Western Days 2010 in Dewey, OK (Washington County). This county had high levels of broadband adoption and decreased sales tax revenues between 1998 and 2008, but the same pattern doesn’t hold across the state. [/imgcontainer]
To test statistically whether increasing broadband rates had any impact on retail sales tax collections, we collected data on a number of variables that could affect sales tax income for all 77 counties in Oklahoma in both 1998 and 2008. We considered features such as the county tax rate, household income levels, age levels, and presence of a Wal-Mart store, converting all variables to 2008 dollars using inflation indices from the Bureau of Labor Statistics.
While the per capita retail sales tax collected increased dramatically over this time period, several other measures also showed increases: county tax rates, the number of Wal-Mart stores per county, the poverty rate, and the level of broadband access.
To determine which of these various increases actually impacted the amount of retail sales tax collections, a technique called “first-difference” regression was used. This technique models the change in a particular variable (in this case, per capita retail sales tax collected) based on changes in other variables that might have impacted it over that time. Only some variables will be statistically linked to the rise in per capita retail sales tax collections. The remaining, insignificant variables can be said to have no effective impact on the retail sales tax collections (for more detail on the study and methodology, see Whitacre (2011)).
So, which variables were determined to be “statistically significant?” Only two:
• Tax Rate
• Poverty Rate
Interestingly, only these two variables (higher rates of taxes and higher poverty levels) had any impacts on the changes in retail sales taxes collected over this period. The higher rates of taxes contributed to higher tax collections, while the higher poverty levels in 2008 decreased the levels of tax collections. Notably, the higher levels of broadband adoption played absolutely no role in the changes that were seen in retail sales tax collections. This pattern held when a separate analysis was done on only non-metropolitan counties in the state.
As local governments face new pressures to pay for their hospitals, police, and other necessary services and infrastructure, the prospect of diminished sales tax collections is ominous. E-commerce has been looked at with suspicion, as a potential threat to local viability. The results shown here suggest that broadband access and Internet sales have thus far not decreased sales tax revenues. Tax collections have not varied in any way based on broadband adoption trends.
Communities can confidently offer educational programs to residents about the benefits of broadband, including encouraging participants to shop on the Internet as they feel comfortable. Some will buy online, and others using the Internet as an auxiliary source of information, will buy products locally. However, e-commerce’s percentage of total retail sales has been trending upward for an entire decade. A continuation of that trend over the next decade could change the findings here.
Brian Whitacre is an Associate Professor and Extension Economist and Lara Brooks an Assistant Extension State Specialist at Oklahoma State University. Their findings were published in Regional Science Policy & Practice, April 2011.