A forum for information and analysis on the Arkansas economy
Evidence continues to accumulate showing that Arkansas’ economy has not been as hard-hit by the COVID-19 pandemic as other parts of the nation. Most of the information to-date has been related to employment. It is clear that declines in payroll employment and spikes in unemployment have not been nearly as large in Arkansas as in the national data. But it appears that one area of the economy has demonstrated an even greater gravity-defying performance: consumer spending.
The latest state revenue report added an underline to the emerging evidence: The Department of Finance and Administration (DF&A) reported on Tuesday that July revenue from Sales and Use taxes was up 14.9% from a year earlier and 16% above forecast. The July revenue corresponds roughly to sales activity in June, a month in which U.S. Retail Sales had recovered from a plunge in April but was still only 1% higher than the previous year. DF&A revenue reports from the two previous months of sales tax receipts had indicated only a small decline in April taxable sales activity and an unexpectedly large increase in May.
The apparent increase in Arkansas taxable sales in June therefore just adds to the evidence that consumer spending in Arkansas has been remarkably resilient to COVID-related shutdowns and the associated economic contraction, relative to both the national averages and expectations. In this article, we break down available data by county and by sector to seek greater insight into the phenomenon.
As we reported a month ago, when county and local sales tax receipts corresponding to spending in April were released, the expected decline in consumer spending was notably absent. Forecasts suggested year-over-year declines on the order of 15%, but an aggregate of taxable sales for Arkansas counties showed little change from a year earlier. Granted, a zero-percent growth rate represented below-trend growth, but U.S. Retail sales were down 19.9% from a year earlier. Some individual counties fared better than others with growth rates ranging from -21.9% to +42.4%. But only 19 counties had negative growth rates and only 42 had growth rates below recent trend rates.
Data corresponding to local taxable sales in May were released in late July, further reinforcing the conclusion that consumer spending declines in the national data were not being reflected in Arkansas sales tax collections. U.S. Retail Sales were recovering, but still down 5.6% from the previous year. Yet only 4 Arkansas counties experienced negative year-over-year growth rates; an aggregate for all 75 counties showed an annual increase of 9.5%; and in over 60 of Arkansas’ 75 counties, taxable sales growth was over 10%. The interactive map below, Figure 1, shows the extent and magnitude of taxable sales growth in May.
National Data vs. Arkansas Taxable Sales
In contrast to Arkansas, data on U.S. Retail Sales and Personal Consumption Expenditures (PCE) indicated dramatic declines in consumer spending during the second quarter. As already mentioned, U.S. Retail Sales and Food Services showed year-over-year declines in March, April and May. By June, Retail Sales had recovered significantly, but were up only 1.1 compared to the previous year. This weakness in retail spending was also reflected in last week’s GDP report: the Bureau of Economic Analysis released data showing that second-quarter PCE was down 10.7% from a year earlier.
As shown in Table 1 and Figure 2 (below), data for Arkansas indicate only modest spending declines. On a year-over-year basis, the largest declines for both U.S. and Arkansas data occurred in April, but the declines in Arkansas were significantly smaller than the 20% drop in U.S. Retail Sales. The first measure for Arkansas, Taxable Sales Including Gasoline (ATSIG), was down only 3.2% in April. An alternative, the sum of county taxable sales (plus the gasoline component of ATSIG), declined only 4.1% from a year earlier.
But as shown in Figure 2, the year-over-year declines for both Arkansas estimates are assessed relative to April 2019, a month in which both measures showed a sharp spike above trend. The time-series illustrated in Figure 2 show that the sharpest month-to-month decline in the Arkansas data occurred in March, not April — and that decline was far smaller than the decline in U.S. Retail Sales. Measuring the change from January through March, U.S. Retail Sales were down 8.6%, while the two measures for Arkansas declined only 5.9% (ATSIG) and 2.8% (county data). U.S. Retail Sales declined another 14.7% from March to April, while the measures for Arkansas showed increases of 4.9% and 1.6%. And as also illustrated in Figure 2, a preliminary estimate of Arkansas Taxable Sales for June ventures into record-high territory.
One notable feature of Figure 2 is the elevated level of Arkansas taxable sales (by either measure) during the second half of 2019. This followed the implementation of Act 822, which mandated out-of-state retailers to collect and remit sales and use taxes for online sales. Using simple linear regression techniques to estimate the trend shift that followed the implementation of Act 822, we estimate the magnitude of the effect to be approximately 3% [an estimate subject to considerable range of uncertainty.] This increase in the second half of 2019 is not literally an increase in taxable sales, but an effective broadening of the tax base to capture a more complete measure of spending. But when comparing year-over-year growth rates from the first half of 2019 through the first half of 2020, the intervening regime shift artificially adds about 3% to measured growth rates. This is part of the explanation for why Arkansas taxable sales have not fallen as sharply as the national retail sales statistics, but it only explains a fraction of the difference.
Another notable feature of Figure 2 is the difference between the two measures for Arkansas in March, the local trough month. As previously mentioned, the statewide measure declined more sharply than the county-based estimate. A likely explanation for this difference involves automobile sales, which are undercounted in our methodology for constructing the county-based measure (given the non-standard formula used to assess local taxes on automobile purchases). We’ll return to that issue, but will exclude autos from the analysis for now.
Retail Sales Components of Arkansas Taxable Sales
We can dig deeper by examining the data for county-level tax collections, which are available from DF&A with monthly breakdowns by sector using the North American Industry Classification System (NAICS). In particular we can examine taxable sales for the specific retail sectors that comprise the U.S. Retail Sales data. In Figure 3 (and Figures 5.1- 5.13, below) we compare U.S. Retail Sales and its components to aggregates constructed using the six largest six counties in Arkansas. Together, these six counties — Pulaski, Benton, Washington, Sebastian, Craighead, and Garland — comprise approximately 50% of total taxable sales in Arkansas. The taxable sales data for this “Arkansas (6)” are seasonally adjusted using implicit seasonal adjustment factors derived from the U.S. Retail Sales data, and gasoline is incorporated into the total by adding 0.5 times the gasoline component of ATSIG. The indexes in the figures are normalized to the first half of 2019, so the scales provide a rough estimate of percent changes from year-ago trend values.
Figure 3 shows total U.S. Retail Trade and Food Services (Excluding Automobiles), along with the six-county aggregate for the corresponding NAICS codes (plus gasoline). Figure 3 indicates that the retail components of Arkansas Taxable sales did, in fact, show a month-to-month decline in April, although the downturn was not outside the range of normal month-to-month variation in the data. Compared to the first half of 2019, the U.S. Retail Sales measure was down 16% in April, while the Arkansas data show a decline of less than 5%.
The fact that this retail-based measure showed a distinct downturn in April suggests that the March trough in Figure 2 was the result of a decline in non-retail taxable sales. This is verified in Figure 4, which shows that a decline in the non-retail component swamped the small increase in the retail component in March, while an increase in non-retail taxable sales April masked the decline in retail spending. Figure 4 also shows that retail spending recovered more sharply in May than the total taxable sales estimates suggest due to another drop in the non-retail component.
A Sector-By-Sector Analysis
In sectors that have been specifically affected by COVID-related shutdowns, Arkansas taxable sales generally show downturns that are aligned with corresponding components of U.S. retail sales, but the magnitudes of the COVID-related contractions are notably smaller. Examples include Furniture & Home Furnishings, Clothing & Accessories, and Food Service & Drinking Places.
Clothing & Accessories experienced the sharpest decline of any sector, with national retail sales falling nearly 90% from the levels of a year earlier. The Arkansas data show a decline of “only” 64%.
Furniture & Home Furnishings declined almost 60% in the national data and dropped about 20% in the Arkansas data.
One of the most prominently hard-hit sectors has been bars and restaurants. Anecdotal information suggests that Arkansas restaurants have been remarkably successful at adapting their operations to the pandemic, with delivery, curbside service and drive-through service substituting for full-service operations. Figure 5.3 demonstrates this in the data: Although sales at Food Service and Drinking Places were declining prior to the onset of the pandemic, they reached a low point in April that was only 27% below the first half of 2019. In the U.S. Retail Sales data, the low point was down over 50%.
Health & Personal Care is another spending category that matches the general pattern of fluctuations in U.S. data, but shows a somewhat sharper decline in Arkansas than the national average. In the U.S. Retail Sales data, Health & Personal Care declined in April to a level only 10% below the previous year, while the decline was over 20% in Arkansas.
(The Health & Personal Care components is one of several that shows a spike in December 2019. This is likely an artifact of end-of-year remittances that were larger than usual, perhaps due to the mid-year implementation of out-of-state sales tax collections.)
Gasoline sales represent another spending component that declined as much or more in Arkansas as the U.S. Figure 5.5 combines county data for taxable sales at gasoline stations with the gasoline sales component of ATSIG, showing a decline of 45%. The U.S. Retail Sales data show a decline of 38%. Unlike spending declines in several other categories, declining gasoline expenditures reflect both falling prices and slowing sales. From January through April, the average retail price of gasoline in Arkansas fell from $2.25 to $1.48, a 34% decline. Meanwhile, the number of gallons of gasoline sold fell by 56%. Although retail sales of gasoline fell sharply through April, the fact that gas is exempt from sales tax in Arkansas means that the decline in this particular spending component had no impact on sales tax receipts by state or local governments.
One other area of spending that showed low sales in April (albeit not a sharp monthly decline) was at Electronics and Appliance stores. The Arkansas data show relatively weak sales toward the end of 2019, with a low point in April that was down 18% from early 2019 levels. Meanwhile, the corresponding component of U.S. Retail Sales was down more than 50%.
Some sectors experienced increases in sales, particularly in the early phase of the pandemic and during the initial economic shutdowns. In both the U.S. and Arkansas data, Food & Beverage store sales surged by approximately 30% in March, and have remained elevated — moreso in Arkansas than the U.S.
Similarly, General Merchandise Stores showed a March spike in both the U.S. and Arkansas data. The subsequent downturn in April and recovery in May still left Arkansas sales well above levels of a year earlier.
Another sector that appears to have benefited from more time spent at home is Building Materials and Garden Equipment & Supplies. In Arkansas, sales in this sector surged to 17% above previous-year levels in March and over 22% in April. In the U.S. data, sales did not rise until May, but had risen to 15% above previous-year levels in May and June.
As a group, Sporting Goods, Hobby, Musical Instrument and Book Stores also seem to have fared relatively well in Arkansas. In the U.S. data April sales were down 45%, but Arkansas showed no significant decline. For both the U.S. and Arkansas measures, pent-up demand appears to be driving a surge in the post-April data.
Another relatively small sector that seems to have thrived in Arkansas during COVID-related shutdowns is Miscellaneous Store Retailers, which has remained above year-ago levels throughout the pandemic and was up 20% as of May. Meanwhile, the national counterpart to this measure dropped 30% in April and is gradually recovering through June.
The sector that shows the largest impact out-of-state sales tax collections is Non-Store Retailers, which includes NAICS sector 4541: Electronic Shopping and Mail-order Houses. As shown in Figure 5.12, Arkansas sales in this sector doubled from the first half of 2019 to the second half of the year (attributable to Act 822) and has shown an increase of similar magnitude during 2020. The increased popularity of online shopping during the pandemic is evident in the U.S. data as well, but it appears that the substitution of internet sales for brick-and-mortar shopping has been more significant in Arkansas than elsewhere.
One final sector to consider is automobile sales — an important component of consumer spending. Sales at Motor Vehicle and Parts Dealers account for about 20% of total U.S. Retail Sales. But because of the quirky way county and local sales taxes are assessed on auto sales in Arkansas ($25 per 1% sales tax rate, regardless of the value of the transaction), the county NAICS data from the Department of Finance and Administration cannot be used to construct a taxable sales measure. And for that matter, used car sales are subject to Arkansas sales tax but are not included in U.S. Retail Sales.
In an effort to approximate the role of auto sales in Arkansas consumer spending, Figure 5.13 compares the U.S. Retail Sales data for Motor Vehicle and Parts Dealers to indexes of two relevant measures that are available using DF&A data. The first measure represents taxable sales at Motor Vehicle and Parts Dealers, excluding the sale of automobiles. It represents a proxy measure for general activity in the automotive sales sector. The second measure, perhaps more relevant, is an index of auto sales that is constructed from the county-level data. It can best be interpreted as an index of the number of vehicles sold (new and used). Both measures show something of a low-point in March, with increases in April and May.
The increases in April and May are consistent with the monthly revenue reports from DF&A, which noted the motor vehicle portion of sales tax receipts was up 25% in June (corresponding to May sales). The latest report noted that motor vehicle sales were up another 12.5% the following month. The low point in March suggests that auto sales were a factor contributing to the drop in total taxable sales in that month, and in the subsequent recovery that masked declines in other spending components. Indeed, In Figure 4 (above) the “non-retail” component technically includes this important motor vehicle sales component.
Conclusions and Explanations
So it turns out that retail spending in Arkansas did, in fact, decline in April — particularly in sectors directly affected by Covid-related shutdowns. But the declines were far smaller than indicated in the U.S. data, and spending growth since April has outpaced the national spending recovery.
Several factor are likely to have contributed to this pattern. First, the COVID-related shutdowns that were implemented in Arkansas were not as restrictive as in many other states, facilitating a somewhat more normal continuation of commerce. Second, the implementation of Act 822 in 2019 increased sales tax receipts for state and local governments, independent of any actual change in consumer spending. But this explains only a portion of the measured year-over-year growth comparisons.
An undeniably important factor has been the role of government support and stimulus payments helping to maintain income and spending. Last week’s U.S. GDP report revealed that personal income from transfer receipts increased by 75% in the second quarter, driving an increase in total personal income of 7.3%. Transfer payments account for a larger share of Arkansas income, so the increase in transfers in the second quarter has likely had an even larger impact here. Forecasts indicate that total personal income in Arkansas (including transfers) increased by 12.5% in the second quarter.
Generally speaking, economists have found that temporary gains in income, like those associated with the stimulus payments and other temporary income support programs, do not lead to one-for-one increases in consumer spending. Rather, some of the windfall is saved or, equivalently, used to pay down household debt. Under the current economic conditions, it appears that the propensity to consume by Arkansas households is high enough that the temporary income gains due to government stimulus programs has been significant enough to not only sustain consumer spending during the shutdowns, but to boost expenditures during the partial re-opening phase.
Looking forward, the prospects for continued strength in consumer spending depend one two factors: First and foremost, the economic outlook overall is dependent on the course of the COVID-19 pandemic and efforts to control it. As noted by Federal Reserve Chair Jerome Powell said last week, “the path of the economy will depend significantly on the course of the virus.” More specifically with regard to consumer spending, the income support provided by government stimulus and support payments has been a crucial factor thus far, and in the absence of a full reopening of a post-pandemic economy, it is likely to be remain an important factor over the next several months. As shown in Figure 6, the recent expiration of some of these programs implies a sharp drop in transfer payments and personal income over the second half of 2020. Congress is presently considering extensions and additional stimulus measures, but the timing and magnitude of those actions are, as yet, unclear.
# # #
The June employment report for metropolitan areas showed a continuing rebound from the job losses of March and April. The table below showed revisions to the nonfarm payroll employment data for May, along with the new numbers for June. The May revisions showed larger employment gains than originally reported in Fort Smith, Hot Springs, and Memphis. Other metro areas saw downward revisions. The employment gains continued into June, with increases ranging from 0.7% in Jonesboro to 3.1% in Hot Springs. Since the Pre-COVID-19 peak in February, employment is down in all of Arkansas’ metro areas, with the magnitude of the declines ranging from -7.2% in Memphis to -2.8% in Pine Bluff.
Unemployment rates generally continued their declines from the previous month, although Memphis saw a slight uptick. The largest decline was in Fort Smith, where the unemployment rate dropped from 11.9% to 7.4%. Hot Springs dropped from 13.8% to 10.6%. Northwest Arkansas continues to have the lowest unemployment rate in the state, dropping to 6.3% in June.
The map below displays a snapshot of unemployment rates by county for June, along with the change from May (not seasonally adjusted data). Unemployment rates were higher in eight counties and unchanged in one. Rates were down in the remaining 66 counties, with 26 counties experiencing declines of more than 1.0% and eight counties seeing declines of more than 2.0%. The highest unemployment rates continue to be generally located in the Delta region, with Chicot County setting the statewide high of 13.6%. Even after an increase of 0.5 percentage points, Arkansas County continues to have one of the lowest unemployment rates in the state at 5.7%. However, after a decline of 0.5 percentage points in June, the lowest rate in the state is now in Madison County (5.2%).
Since our last forecast update six week ago (June 10), incoming data have continued to show a more rapid recovery from the COVID-19 shutdowns than previously expected. Nonfarm payroll employment bounced back with increases of 1.9% and 2.0% in May and June, respectively, leaving total employment down only 5.7% from February. The unemployment rate declined from its April peak of 10.8% to a rate of 8.0% in June. Meanwhile, the pace of government stimulus payments has further boosted personal incomes, and data on local sales tax collections indicates somewhat smaller declines in consumer spending than earlier projected.
In this updated forecast report, we present new projections for the Arkansas economy in light of recent developments, incorporating newly-published data. The forecast is based on the latest update to the national outlook from IHS Markit, dated July 8. The July IHS forecast projects second-quarter GDP growth falling at an annual rate of 35.5% – a somewhat smaller drop than expected the previous month. The national unemployment rate peaked at 13.4% in the second quarter, and is expected to decline gradually over the next several quarters, reaching single-digits by the end of this year and down to 6% by the end of 2021.
As with our previous forecasts for the Arkansas economy, we have applied the recent IHS national forecast changes to the model for Arkansas, adjusting the components of each aggregate separately in order to capture some of the unique characteristics of the Arkansas economy. Our forecast is also informed by updates to the IHS Arkansas model published in June and July.
The following sections present our latest post-COVID-19 forecasts for Arkansas, comparing the latest projections to previous forecasts. The figures presented here compare the current (July) and previous (June) forecasts, relative to the pre-COVID-19 baseline from February.
The most significant revision to the forecast is, once again, the outlook for employment. In June we had expected employment declines to moderate, with job losses continuing into the third quarter. The sharp rebounds that we saw in May and June are now baked into the forecast for the third quarter, implying employment growth of 4.1% relative to the second quarter. Employment is expected to recover to pre-pandemic levels by mid-2021.
The monthly household survey on which the unemployment rate is based continues to encounter logistical difficulties and is also plagued by problems with categorizing workers who have been temporarily idled by COVID-19 shutdowns. Since March, the BLS has noted that many workers affected by COVID-19 shutdowns were incorrectly classified as “employed but absent from work” instead of “unemployed on temporary layoff.” The magnitude of this misclassification appears to be diminishing over time, but there is still greater-than-usual uncertainty about the unemployment statistics.
Measurement problems notwithstanding, Figure 2 shows the latest outlook for Arkansas unemployment. Just a month ago, the forecast indicated unemployment would average over 10% in both the second and third quarters. The preliminary data indicate that the unemployment rate peaked at 9.5% in the second quarter and is now expected to decline gradually over time. The unemployment rate should drop below 6% sometime in 2021, but is unlikely to return to full-employment levels until sometime the following year.
Estimates of income from government stimulus payments continue to raise personal income estimates for the second quarter. It is now estimated that second-quarter transfer payments (which include special tax rebates, expanded unemployment insurance benefits, and other social safety-net programs) will be 66% higher than in the first quarter. The higher estimates for transfer payments reflect both the magnitude and speed with which payments are being sent out. Congress is currently considering legislation to extend some of the programs established by the CARES Act, but the current forecast assumes that programs will expire as scheduled. Consequently, while it is likely that elevated transfer payments will continue to raise incomes in the second half of 2020, the current forecast maintains the assumption that transfer payments will drop sharply in the third quarter.
For total personal income, the current forecast estimates that Arkansas Personal Income in the second quarter was 14.6% higher than a year earlier. With transfer payments expected to drop off in the third quarter, income will still be up 4% on a year-over-year basis.
Figures 3a and 3b show how the total income forecast is reflected in expectations for transfer income and non-transfer income. Net of transfer payments, personal income is now expected to show a decline of 4.3% in the second quarter—and improvement over the 7.8% decline in the June forecast. The July forecast also reflects newly available data for the first quarter, showing that non-transfer income dropped 0.2% instead of the 0.4% increase that was expected.
Despite the boost to incomes from expected transfer payments, we are still anticipating that the data will show a sharp decline in Personal Consumption Expenditures (PCE) in the second quarter. Nevertheless, recent data on sales tax collections suggest that the decline in consumer spending may be overstated. If federal stimulus programs are extended beyond current-law deadlines, recovery in the third quarter might be more pronounced.
The July forecast indicates that PCE declined 10.9% in the second quarter, compared to the 12.2% decline expected in the June forecast.
Table 1 compares the July forecast relative to the May and June forecasts for PCE, focusing on the two-quarter impact (2019:Q4-2020:Q2) of the COVID-19-related shutdowns. Projections for many of the components have changed little since last month, while others reflect specific sectoral information collected by IHS forecasters.
The largest revisions are in Durable Goods, for which data on both the national and state level indicate stronger-than-expected demand. This is particularly true for Motor Vehicles and Parts, for which early projections turned out to be far too pessimistic. The decline in nondurable goods is now expected to be somewhat smaller than the June forecast, while the estimate of spending on services is now slightly lower than in the June forecast.
Figure 5 shows index values (relative to 2019:Q4=100), illustrating the relative paths for total PCE, Durable Goods, Nondurable Goods (less Gasoline & Other Energy) and Services. When gasoline expenditures are excluded, nondurable goods purchases are expected to remain relatively stable and increasing over time. Subtracting Food and Beverages Purchased for Off-Premises Consumption, the remaining components are now expected to decline by 7%. Spending on services is expected to decline by 15%. In the June forecast, durable goods purchase were expected to show the largest decline of any component, over 19%. The initial decline is now only expected to be only 10%, but growth is likely to remain relatively sluggish in coming quarters. In the current outlook, it is services spending—particularly in those sectors directly related to COVID-19 shutdowns—that show the most significant decline.
Implications for Local Sales Tax Collections
One of the practical applications of the consumer spending forecast is evaluation of how sales tax receipts by city and county governments might be affected. To assess the sales tax implications, we have been tracking a composite measure that is intended to mimic the taxable sales base. This taxable sales proxy—shown in Figure 6—includes all Durable and Nondurable goods less Gasoline, plus Recreation Services and Food Services & Accommodation.
The outlook for taxable sales has improved in this month’s forecast, for the third consecutive month. The June forecast had shown a decline of 16.1% (from 2019:Q4 to 2020:Q2). The latest forecast shaves that loss down to 12.9%. Reflecting the more rapid and robust recovery of employment and income that is now projected, the taxable sales base is now expected to rebound more sharply than in either the April or May forecast.
# # #
The state employment report for June showed continued recovery from the sharp job losses registered in April during the peak of pandemic-related economic shutdowns. The Arkansas unemployment rate dropped from a revised 9.6% in May to 8.0% in June. The national unemployment rate declined from 13.3% to 11.1%. The household survey indicated a decline in the number of unemployed Arkansans of nearly 24,000 to 105,338. The number of employed declined as well (down 9,300), resulting in a sharp downturn in the measured labor force. The Arkansas labor force participation rate has dropped 2.4 percentage points from 58.2% in January and February to 55.8% in June. The U.S. labor force participation rate declined 1.9 percentage points (from 63.4% to 61.5%) over the same period.
Arkansas nonfarm payroll employment rose by 23,700 in June, an increase of 2.0%. The June increase was proportionately not quite as large as the national total, which showed an increase of 3.6%. From February through June, the net decline in Arkansas payroll employment was 5.7%, while the U.S. total showed a decline of 9.6%.
Employment gains continued to be concentrated in the sectors that were most effected by pandemic-related shutdowns. Employment in Retail Trade increased by 10,300 in May and June, almost completely offsetting the April decline. Leisure and Hospitality Services added 8,500 jobs in June, but remains far below levels of a year ago. Significant gains were also registered in June for Professional and Business Services (particularly in Administrative and Support Services) and in Other Services.
# # #
Seasonally adjusted data for Arkansas nonfarm payroll employment, reported in a format consistent with the monthly news release from the Arkansas Division of Workforce Services, can be found here: Table-Seasonally Adjusted NFPE.
In the latest report on state-level GDP, the U.S. Bureau of Economic Analysis reported that Arkansas’ economy contracted at a 5.0% annual rate in the first quarter, the same rate as the U.S. That was also the median growth rate among states. GDP declined in all 50 states, with growth rates ranging from -1.3% in Nebraska to -8.2% in New York and Nevada.
Over the course of 2018 and 2019, Arkansas GDP growth averaged 1.9% compared to the U.S. average growth rate of 2.4%. Arkansas growth rate had exceeded that of the U.S. in the second half of 2019. Accordingly, year-over-year increase for Arkansas is slightly higher than the U.S.: +0.5% versus +0.3%.
The breakdown of GDP by sector shows that although Arkansas contracted at the same rate as the U.S., the composition of the decline was somewhat different. Arkansas benefited from relatively strong growth in Construction and Management of Companies and Enterprises. Manufacturing contributed more to Arkansas decline than to the U.S. for two reasons: Manufacturing represents a larger share of the state’s output and both durable goods and nondurable goods manufacturing declined more sharply in Arkansas. In sectors considered most-affected by pandemic-related shutdowns (e.g. Retail Trade, Health Care, Accommodation & Food Services, etc.), Arkansas’ declines tended to be smaller than the national average.
The May employment report for metropolitan areas showed slight employment gains in all of Arkansas’ metro areas, offsetting a portion of the sharp declines in April. The table below showed revisions to the nonfarm payroll employment data for April along with the new numbers for May. The preliminary numbers for April had originally shown that all of the metro areas contracted as much or more than the statewide average, suggesting that non-metro areas in the state fared better than average. After revisions, there is more variation among the metros. Employment totals for Fort Smith, Jonesboro, Pine Bluff and Texarkana were revised to show smaller losses April, while the magnitude of the statewide decline was revised from -8.0% to -8.6%. The preliminary numbers for May show rebounds in all metro areas, with the size of the rebound generally related to the size of the April decline. Hot Springs bounced back from a 14.9% loss to register a gain of 6.3% in May. Jonesboro’s revised April job losses were 6.9%, with a May rebound of only 0.9%. Since February, cumulative changes in nonfarm payroll employment range from -3.7% in Pine bluff to -10.1% in Hot Springs.
Unemployment rates also showed a partial reversal from April, with rates declining in all of Arkansas’ metro areas. Texarkana saw the largest decline (2.9 percentage points) followed closely by Hot Springs (2.6 percentage points). Pine Bluff showed the smallest decline in unemployment (-0.4 percentage points) but had also experienced the smallest April increase among Arkansas’ metro areas.
The pattern of reversals from April’s unemployment rate spike generally carried over to the county level as well. Unemployment rates were unchanged in four counties and were slightly higher in 5 counties. The other 66 counties saw declining unemployment. The interactive map below shows unemployment rates for May, along with the change from April’s (revised) rates. (Note that the data are not seasonally adjusted. Statewide, the seasonal change in the unemployment rate from April to May averages +0.3%)
The Arkansas Department of Finance and Administration has published new data on local sales tax collections paid to counties and municipalities in June. There is generally a one month lag between actual sales and the collection of sales tax and an additional month for the distribution to be made to the local governments. Consequently, the sales taxes remitted to the counties and cities in June largely reflect sales in April – the month in which COVID-19 related shutdowns are likely to have had their largest impact on economic activity.
We had projected declines on the order of 15% on a year-over-year basis, based on the estimated impact of COVID shutdowns on retail sales. As it turns out, most counties in Arkansas fared better than that, albeit with significant diversity among counties. The interactive map below shows the percentage changes in sales tax receipts by county between April 2019 and April 2020 (expressed in terms of taxable sales as an adjustment for tax-rate changes). The largest decline was in Carroll County (-21.9%), while the largest increase was Calhoun (+42.4%).
There are several possible explanations for the better-than-expected sales tax figures. First, it might be the case that support from government transfer programs helped maintain consumer spending, in spite of income losses due to layoffs and business closures. Second, sales tax in Arkansas is collected on more than just final retail sales to consumers; tax collections on other business-to-business transactions might have mitigated the effects of declines in retail sales. Third, the recent change in the law that required out-of-state retailers to collect and remit sales taxes from Arkansas residents began in July 2019, so that the year-over-year comparison for April includes an expansion of the tax base.
A preliminary analysis of sales tax collections by sector for Pulaski County suggests that all of these factors contributed to the April 2019-20 comparison. Tax collections in services categories showed the expected declines; for example, Arts, Entertainment, and Recreation was down 45.6% and Accommodation and Food Services were down 41.9%. In contrast, Retail Trade was down only 2.0%, with large losses in some sub-sectors (e.g. Furniture stores down 29%) offset by others (e.g. Beer, Wine, and Liquor Stores up 40%). Patterns of consumer spending were affected, but overall consumer spending appears to have fallen less than anticipated. Increases in sales tax collections were particularly strong in the non-retail sectors; Collections from Manufacturing (primarily from Iron and Steel manufacturing) were up 234% and taxes related to Construction activity were up 242%.
There is also evidence that the collection of sales tax from out-of-state vendors played a significant role in sustaining sales tax collections. In the Pulaski County data, sales tax collections from Sector 4541, Electronic Shopping and Mail-Order Houses, was up 205%. Without this component, the total for Retail Trade would have declined 8.6% instead of 2.0%.
In order to approximate an impact of internet sales tax collections on the county statistics, the interactive map below compares the year-over-year growth rate for April, relative to a trend growth rate that is calculated as the average of year-over-year changes from July 2019 through February 2020 (corresponding the the period since out-of-state vendors were required by law to collect and remit taxes). With this adjustment, the relative performances of counties changes little, but the interpretation of the numbers is altered. For example, the raw April-April growth rate for Garland County was -9.2%. Compared to a trend rate of growth averaging 10.2%, the April growth rate is 19.5% below trend (after rounding).
One other factor that should be considered: The correspondence between sales in April and taxes remitted to the counties in June is a general, but not exact relationship. Not all tax reporters submit monthly, and the tax receipts are reported for the month in which they are received. Any extraordinary effects of reduced taxable sales in April are likely to be reflected, in part, in subsequent monthly reports of tax remittances.
Data for city and county tax collections, broken down by NAICS-code sectors, are available for all tax jurisdictions in Arkansas. We will continue to monitor and analyze these data in our continuing efforts to gauge the impact of COVID-19 (and efforts to prevent its spread) on the economy of Arkansas.
Note: Monroe County has no countywide sales tax and Saline County initiated one in April 2019. For these two counties, the taxable sales figures reported are aggregates of the cities with sales taxes in those counties.
The Bureau of Economic Analysis released the state-level personal income report for the first quarter of 2020 this morning. The BEA noted that this was the first-quarter to be “impacted by the response to the spread of COVID-19, as governments issued ‘stay-at-home’ orders” in the latter part of March.
The report shows that Personal Income in Arkansas rose at a 1.9% annual rate in the first quarter, somewhat slower than the U.S. rate of 2.3%. Nationwide, growth rates ranged ranged from 4.9% in New Mexico to -0.3% in Michigan
Arkansas’ growth rate for the fourth quarter of 2019 was revised down slightly from 2.7% to 2.5%. Over the past four quarters personal income in Arkansas has increased 3.3%, while U.S. growth has been 3.2%.
Arkansas growth rate in the first quarter was again buffeted by swings in Farm Income, which declined at a 68.5% annual rate, subtracting 0.9 percentage points from the overall growth rate. The decline in Proprietors’ Income was also driven by Farm Proprietors’ component. Nonfarm Proprietors’ Income was up at a 3.7% rate. The BLS report noted that “increases in transfer receipts and property income (dividends, interest, and rent) contributed to personal income growth in all states and the nation.” In Arkansas property income accounted for 0.6 percentage points of the states total personal income growth; transfer payments accounted for 1.4 percentage points. Net earnings (which excludes contributions for government social insurance programs and is adjusted for state of residence) accounted for a tiny net decline (-0.1 percentage points) in total income.
The BEA noted that increases in transfer payments “reflected a 1.6 percent cost of living adjustment for social security recipients; an increase in state unemployment insurance compensation; and an increase in refundable tax credits.”
The table below shows a breakdown of total earnings by sector, expressed as contribution to total personal income growth. In addition to losses from Farm earnings, Arkansas saw losses in Mining, Durable Goods Manufacturing, along with several service-sector categories. Earnings in Health Care, Accommodation and Food Services, and other service sectors experienced earnings declines, but of smaller magnitude than the U.S. totals.
Covering the first quarter, today’s report encompassed only the very early weeks of the COVID-19 pandemic and the business shut-downs that it engendered. The BEA took special steps to assure that the full impacts were reflected: “BEA used both March and April CES state level data to better capture business closures and job losses that occurred in March. BEA also used weekly state level Unemployment Insurance (UI) claims data to inform its 2020 Q1 state estimates of wages and salaries.” A fuller accounting will likely be embedded in the report for the second quarter.
Last month, the April employment report for Arkansas was not as bad as expected. The May report is better than expected.
The unemployment rate dropped from a revised 10.8% in April to 9.5% in May. The number employed rebounded sharply from its April decline, up 46,378. The number of unemployed dropped by 12,663 from 140,898 to 128,235. The labor force total continues to fluctuate widely, increasing by 33,715 in May after declining by 76,788 in April.
The drop in unemployment reported in the household survey is certainly welcome news, but it is somewhat at odds with the unemployment insurance claims data. For both Arkansas and the U.S., continuing claims for unemployment insurance were higher in May than in April, and both continuing claims and initial claims remain elevated as of the latest data released yesterday. Expressed as a percent of the labor force, the continuing claims data suggest insured unemployment rates of approximately 14% for the U.S. and 10% for Arkansas.
The elevated unemployment insurance claims levels suggest that there is a great deal of churning going on in labor markets, complicating the interpretation of the unemployment data. In addition, there continue to be some survey problems in the proper classification of “employed” versus “unemployed” for persons displaced by COVID-related shutdowns. But the clear gains employment (reinforced by the payroll data), indicate that the labor market improved in May, and that Arkansas has thus far fared better than the rest of the nation, on average.
Nonfarm payroll employment jumped 21,400 in May, bouncing back somewhat from a revised 109,200 decline in April (seasonally adjusted). The April report had shown proportionally smaller declines for Arkansas than in the U.S. data, particularly in sectors directly affected by COVID-related shutdowns (e.g. Retail Trade and Accommodations & Food Services). Consequently, we had not expected to see as sharp a rebound in May as reported in the national data. As it turns out, the May rebound in Arkansas was a 1.8% jump, nearly equal to the 1.9% increase for the U.S. From February through May, Arkansas payroll employment has shown a net decline of 7.6%, compared to 12.8% in the U.S. data.
A breakdown by sector shows that the largest employment gains were in Leisure & Hospitality Services, Education & Health Services, and Retail Trade, each of which suffered particularly sharp declines in April. Not all sectors saw increases in May: Employment in Wholesale Trade dropped slightly, and Manufacturing employment slid deeper into negative territory. Government employment declined by 4,400, due to employment declines at educational institutions. This decline is not unusual, but it typically shows up in June and July. Compared to a year earlier, payroll employment in Arkansas was down 91,000 jobs.
# # #
Seasonally adjusted data for Arkansas nonfarm payroll employment, reported in a format consistent with the monthly news release from the Arkansas Division of Workforce Services, can be found here: Table-Seasonally Adjusted NFPE.
As the economic impact of the COVID-19 pandemic continues to unfold, one constant is the rapid pace at which the situation is developing. Just one month ago, our May forecast came out in the wake of the April employment report, which showed U.S. payroll employment plummeting by over 20 million and the unemployment rate rising to 14.7%. The ongoing accelerated pace of initial claims for unemployment insurance suggested that labor markets would continue to decline, both nationally and here in Arkansas.
Over the past month, however, there have been emerging bits of information that suggest that Arkansas has not been as hard-hit as many other parts of the country. While initial claims for unemployment insurance remain high, continuing claims (insured unemployment) seemed to have clearly plateaued at around 120 thousand, implying an insured unemployment rate of about 10%. The U.S. insured unemployment rate has shown some signs of coming down from a peak in early May, but remained above 14% near the end of the month.
The release of the state-level employment report for April provided further evidence of Arkansas’ relative resilience to the economic downturn. If every sector in the Arkansas economy had suffered the same proportionate losses as shown in the U.S. employment report, we would have expected a decline of 160 thousand jobs in April. Instead, the decline was “only” 100 thousand. It was telling that the sectors in which Arkansas outperformed the national economy included some of those most-impacted by COVID-related economic shutdowns; for example, both Retail Trade and Leisure & Hospitality contracted far less than the national average. Construction and Manufacturing also experienced smaller declines in Arkansas than in the national data.
Another piece of information came from the state revenue report for May, which showed Sales and Use Tax collections down only 2.8% from a year earlier. That was 8.5% above the forecast of the Arkansas Department of Finance and Administration and similarly above our expectations.
But the most paradigm-disrupting development over the past month was the unexpected rebound of U.S. labor markets that was announced last Friday (June 5). The May employment report showed a one-month increase in payroll employment of approximately 2.5 million and a decline in the official unemployment rate to 13.3%. The increase in payroll employment was the largest one-month gain ever recorded for the series and was the largest monthly percentage increase since WWII. In the context of the 22 million decline in employment over the previous two months, the May rebound might be characterized as a “dead-cat bounce.” Nevertheless, it showed that the limited re-opening of the economy around the country was generating a more rapid and robust resurgence of economic activity than most economists had expected.
In this updated forecast report, we present new projections for the Arkansas economy in light of recent developments. The forecast is based on the latest update to the national outlook from IHS Markit, originally dated June 4, but updated on June 8 to account for the surprising U.S. employment report. The June IHS forecast projects second-quarter GDP growth falling at an annual rate of 42% — down from the 36.5% rate of decline forecast a month earlier. National unemployment is now projected to peak earlier at only 13.4%, with the peak now coming sooner (in the second quarter rather than the fourth).
As with our previous forecasts for the Arkansas economy, we have applied the recent IHS national forecast changes to the model for Arkansas, adjusting the components of each aggregate separately in order to capture some of the unique characteristics of the Arkansas economy. Our forecast is also informed by a recent update to the IHS Arkansas model that reflected the assumptions in the previous (May) national model. In addition, this forecast update is adjusted along several margins to reflect the relatively-strong performance of the Arkansas economy. We take explicit account of the effects of higher earnings and spending that are likely to accompany the improved employment outlook, adjusting the model-generated forecasts to match emerging data for the state.
The following sections present our latest post-COVID-19 forecasts for Arkansas, comparing the latest projections to previous forecasts.
The most significant revision to the forecast is, once again, the outlook for employment. A month ago, we expected payroll employment to decline through May or June, with recovery slowly taking hold in a partially-reopened economy. The rebound of U.S. employment in May changes that outlook dramatically. Because the Arkansas economy has been proportionately less affected by employment declines in March and April, a significant May rebound is considered unlikely. Nevertheless, the flow of job-losses is expected to have abated.
The latest forecast indicates total job losses of 110,000 in the second quarter, a drop of 8.6% from the first quarter. The May forecast had projected job declines nearly twice that size. A likelihood remains that we might see further declines in the third-quarter as firms adjust to the limited reopening of the economy. Gradual recovery and steady positive job-growth is now expected before the end of the year.
Even in the best of times, the unemployment rate can be a tricky measure to interpret. Based on a nationwide survey, the rate presents the number of unemployed as a percent of the labor force, where movement into and out of the labor force complicate matters. In the present setting, interpreting the underlying survey results has also been problematic. In March, April and May, the Bureau of Labor Statistics noted that many workers affected by COVID-19 shutdowns were incorrectly classified as “employed but absent from work” instead of “unemployed on temporary layoff.” The BLS reports the survey results as the official statistics, but has provided estimates of how this misclassification might have affected the reported unemployment rate. As shown in Table 1, correcting for the misclassification might add as much as 3 to 5 percentage points to the reported rate. The BLS has reported that the error is not specific to any state or region, so it is unclear how it has affected the measurement of Arkansas’ unemployment rate.
Measurement problems notwithstanding, Figure 4 shows the latest outlook for unemployment based on the typical relationship between household employment and payroll employment. Just a month ago, the forecast indicated unemployment in the range of 16 to 17% through the end of 2020. The latest forecast shows unemployment remaining at around 10.3% through the summer before starting to decline in the fourth quarter.
The latest forecast for personal income continues to reflect a combination of lower earnings, offset by a deluge of government transfer payments (which include special tax rebates and expanded unemployment insurance benefits). As shown in Figure 5, the effects of transfer payments imply a surge personal income in the second and third quarters, dropping off sharply after that.
Figures 5a and 5b show how transfer income and non-transfer income are affecting overall personal income. Reflecting higher expected employment (and wages), the June forecast for personal income less transfer payments is expected to decline by only 7.8%. In May the forecast decline was 10.3%. Meanwhile, the timely distribution of tax rebate payments has increased total transfer payments expected in the second quarter, in part by moving them up from the third quarter. In the absence of further policy actions, the transfer payments authorized under the CARES Act are expected to diminish in the fourth quarter.
Despite the boost to incomes from expected transfer payments, we are still anticipating a sharp decline in Personal Consumption Expenditures (PCE), particularly in the second quarter. The combination of business closures and reduced non-transfer income is expected to suppress consumer spending temporarily. Ongoing economic weakness is expected to keep spending from accelerating to pre-COVID levels for some time. Although the surge in transfer income could buffer consumer spending from even larger declines, economic theory and evidence suggests that one-time payments result in relatively small boosts to current spending, with significant portions of the payments being saved or (equivalently) used to pay down household debt.
As shown in Figure 6, the latest forecast suggests a somewhat smaller decline in PCE than projected a month ago. The outlook now includes a slightly more rapid pace of recovery, reflecting improved expectations regarding employment and income. Spending in the second quarter is now projected to decline by 13.4%, slightly less than the 14.8% decline predicted in the May forecast.
Table 2 compares the May and June forecasts for PCE, focusing on the two-quarter impact (2019:Q4-2020:Q2) of the COVID-19-related shutdowns. Projections for many of the components have changed little since last month, while others reflect specific sectoral information collected by IHS forecasters. Each of the series was adjusted to account for the higher income profile embedded in the June forecast.
The largest revision are in Durable Goods, particularly the Motor Vehicles & Parts component, for which data on both the national and state level indicate stronger-than-expected demand. The decline in nondurable goods is now expected to be somewhat larger, as some of the strength in Food & Beverages Purchased for Off-Premises Consumption (a.k.a. groceries) is expected to dissipate as the economy opens further. Spending on services is still expected to show sharp declines, although the magnitude of the drop in Food Services and Accommodations is now somewhat smaller.
Figure 7 shows index values (relative to 2019:Q4=100), illustrating the relative paths for total PCE, Durable Goods, Nondurable Goods (less Gasoline & Other Energy) and Services. When gasoline expenditures are excluded, nondurable goods purchases are expected to remain relatively stable throughout the forecast horizon. But subtracting Food and Beverages Purchased for Off-Premises Consumption, the remaining components are expected to decline by about 10%. Driven by reduced spending on Recreation and Food Services & Accommodation, spending on services is expected to decline by 15%. The largest declines are in durable goods, which falls by 19%.
Implications for Local Sales Tax Collections
One of the practical applications of the consumer spending forecast is evaluation of how sales tax receipts by city and county governments might be affected. To assess the sales tax implications, we have been tracking a composite measure that is intended to mimic the taxable sales base. This taxable sales proxy—shown in Figure 8—includes all Durable and Nondurable goods less Gasoline, plus Recreation Services and Food Services & Accommodation.
The outlook for taxable sales has improved in this month’s forecast, for the second consecutive month. The April forecast had shown a decline of 18.3% from 2019:Q4 through 2020:Q2. In May, the projected decline was 16.7%. The latest forecast shaves that loss down to 16.1%. Reflecting the more rapid and robust recovery of employment and income that is now projected, the taxable sales base is now expected to rebound more sharply than in either the April or May forecast.
# # #
 A dead-cat bounce is a term used in finance to refer to a small recovery in the price of an asset after a period of sharp decline. The name comes from the idea that even a dead cat will bounce if it falls far enough or fast enough. As a cat owner, the Arkansas Economist intends no disparagement of cats.