Accommodation employment growth and volatility: Welcome aboard for a rocky ride

COVID-19

This study was conceptualized before the term coronavirus was in most of our vocabularies, and the data collected, encompassing a 5-year period from 2014 to 2018, reflect a time when what we are experiencing as this research note is written was unimaginable. But, as noted below, the research provides findings that may prove to be even more relevant once we emerge from the pandemic and communities seek the best avenues to spur recovery than when the research was conceived.

Introduction

How stable is employment in the tourism industry? How has the industry grown from an employment perspective versus employment across the broad economy? This US-based study addresses these questions, using accommodation sector employment as surrogate for the broader tourism industry. Why accommodations? Tourism, though an industry we all conceptually understand, lacks a governmental—Bureau of Labor Statistics (BLS)—North American Industry Classification System (NAICS) classification, and as such, tourism employment data per se are not available. This makes sense. Food services, though with an NAICS classification of its own, represent an important subsector of the tourism industry. However, restaurants, we all recognize, derive significant revenue from nontourists. Attractions and retail share this issue. Accommodations, however, represent a nearly pure tourism employer, providing an available and effective barometer for the broader tourism “industry,” and is thus this study’s focus.

Literature

Timo (1993: 33) stated, over 25 years ago, and certainly true today, that “considerable policy interest has been expressed in the employment generating potential of tourism.” But there is more to employment than simply the number employed. Specifically of interest to the current study is the issue of employment volatility. It is important, however, to recognize the distinct difference between employment volatility and employee turnover, with turnover, over 70% in the US lodging industry (Haussman, 2016), resulting in “a significant loss of investment in human capital, training, and quality” for the industry (Davidson et al., 2010: 452). Volatility, conversely, does not consider who is employed for how long but rather measures the variance of total number of employees in the sector, over time.

Burns (1960) posited decades ago that the US’s shifting economy, moving from manufacturing to services, would mitigate future employment fluctuations. Forty years later, Warnock and Warnock (2000) tested Burns’ projection, finding economy-wide numbers failed to properly tell the story, and that only by disaggregating the data by sector could one get a sense of the volatility, or lack thereof, of national employment. Using post-WWII data (1946–1996), Warnock and Warnock (2000: 9) determined: “In general, employment volatility in service-producing sectors…is lower than in good-producing sectors.” These authors further noted the reduced volatility of manufacturing employment, which they pinned on two structural changes not anticipated by Burns: improved Federal Reserve monetary policy and the adoption of “Just-in-Time Inventory Controls” by manufacturers. As a result, while overall volatility did in fact decrease as Burns had predicted, what Burns had not foreseen was that manufacturing would primarily be responsible for the increased stability. Warnock and Warnock (2000: 12) concluded: “While Arthur Burns was to some extent correct…we present evidence suggesting that a key catalyst is the declining employment volatility in the manufacturing sector itself” and the segment’s structural advantage, from an employment stability perspective, versus services [which, of course, includes tourism].”

A more recent, albeit limited, study by Dufrene and White (2007), which employed Berry and Blackwell’s (2005) model, considered employment volatility of industry sectors across Indiana. These included a broad category labeled Leisure and Hospitality (combining NAICS-codes 71 (arts, entertainment, and recreation) and 72 (accommodations and food services)). Dufrene and White (2007: 9) determined that leisure and hospitality exhibited high volatility in all metro areas studied. Several enjoyed annual growth rates exceeding the national growth in employment, thus classifying as “acceptable.” Most, however, experienced positive or neutral annual growth rates, albeit with high volatility, “clearly reveal[ing] a drawback of leisure and hospitality” as an industry for communities to pursue.

Given Berry and Blackwell’s (2005) less than stellar classification of industries similar in nature to tourism, Warnock and Warnock’s (2000) preferential view toward manufacturing versus services, and Dufrene and White’s (2007) lack of support for leisure and hospitality, why would a community choose to target, from an employment perspective, hospitality/tourism?

Scheyvens (2012) argued that economies that expand tourism positively impact their monetary injections through tourism receipts, economic growth, and job creation. Williams and Shaw (1988: 93) observed that while the preponderance of hospitality employment is at the “lower end of the skills continuum,” positions, such as servers, front-desk agents, and housekeepers, are virtually impossible to automate and thus highly stable, with many positions encompassing supervisory duties and offering tangible career progression. Another advantage, per the United Nations (UNDP, 2011), is that tourism employs more women and young people than most industries. Finally, as noted by Dogru and Bulut (2018: 425), the “bidirectional causality between growth in tourism receipts and economic growth suggests that…tourism development stimulates economic growth and vice versa.”

Given the lacuna created by the mixed literature, much strongly in support of tourism as an employment positive, with others leery of tourism’s volatility, it is evident that a comprehensive analysis of hospitality employment data can provide guidance for governmental officials and industry leaders as they consider investing in tourism development in their communities. To actuate such research, the following hypotheses are tested:

H1: Accommodation sector employment growth exceeds national employment growth.

Thus, the hypothesized conclusion, based on Berry and Blackwell’s (2005) model, is that the accommodation sector, and therefore, by extension tourism, is an “Acceptable Target” (high growth and high volatility) for community development initiatives.

Method

Carlino et al. (2013: 522) argue that while much research “examines the volatility pattern of aggregate economic variables…there are few studies that use state-level data to better understand the factors driving fluctuations in volatility.” It was felt that the use of county-level data herein, as discussed below, is better still, providing useful information for tourism planners at a governmental level, where tourism policy initiatives are often implemented and where results can be effectively measured.

There are approximately 3200 US counties. Of these, the government classifies 315 as “large.” With several exceptions, these have populations >150,000 residents. Limiting research to counties with substantial populations eliminates the swings smaller counties would perhaps experience from the addition or loss of even a single hotel property.

Using BLS websites, general employment statistics for the 315 large counties were extracted for the years 2014–2018, ¹ with these, then coupled with the counties’ lodging employment statistics ² for the period. These data were then analyzed to determine employment growth and annual volatility for each county over the tested period. Additional analyses considered county population as well as relative importance of tourism. As the dataset encompasses all US counties over the population threshold, the data reflect, without any issue related to sampling error, a 5-year employment census of what could be described as nonrural US counties.

Findings

The primary analysis compared the growth rate and volatility of lodging employment versus total employment. It was noted that total employment in the tested counties had grown an average 7.4% over the 5-year period…considerably less than the accommodation sector’s 9.3% growth rate over the period. Employment volatility, measured as the average absolute annual fluctuation over the tested period for each county, was also found to be substantially higher for the lodging sector (mean = 4.9%) than was experienced by these counties’ total employment base (mean = 1.9%). These results, reflecting high growth accompanied by high volatility, align with the hypothesized outcomes, comfortably classifying accommodations and thus tourism, per the Berry and Blackwell (2005) model, as an “Acceptable Target” industry for communities to pursue. To further explore these results, counties were segmented by population into three categories. Regardless of population, the above results remained consistent, with NAICS-721 employment growth higher than the county’s overall employment growth, accompanied by higher than average employment volatility. When counties were then segmented based on their reliance upon tourism, determined by the accommodation sector’s share of county employment, high growth accompanied by high volatility was once again noted for each segment. (Results and segmentation criteria are reflected in Table 1.)

OPEN IN VIEWER

Table 1.

Employment growth and volatility for large US counties, 2014–2018.

	Mean 5-year employment growtha		Mean annual volatilityb
	All sectors	NAICS 721	All sectors	NAICS 721
All “large” counties (n = 315)	7.4%	9.3%	1.9%	4.9%
Small “large” counties (n = 186) Population < 500,000	6.7%	8.4%	1.8%	5.3%
Mid-size counties (n = 89) Population 500,000–1,000,000	7.4%	10.1%	1.8%	4.8%
Largest counties (n = 40) Population > 1,000,000	10.6%	12.0%	2.6%	3.5%
Low-tourism-employment (<1%) (n = 192)	7.2%	8.4%	1.8%	5.3%
Mid-tourism-employment (1–2%) (n = 92)	7.3%	10.3%	1.9%	4.3%
High-tourism-employment (>2%) (n = 31)	8.9%	12.3%	2.3%	4.5%

^a Employment growth measures the gross change of employment over the 5-year test-period.

^b Volatility is measured as the average absolute annual employment fluctuation over the 5-year test-period.

Conclusion

Based on the Berry and Blackwell (2005) model, we thus learn, for the US’s 315 “large” counties over a 5-year test period, that tourism, as hypothesized, should be viewed as an “acceptable” industry sector to pursue. This remained true when counties were segmented both by population size and tourism reliance. It would have been a more favorable result to have learned that tourism was a “desired” sector, but until the industry can reduce its volatility, such a classification will remain elusive.

The employment growth reflected in the above findings is indeed encouraging, but the literature provides warnings that the pursuit of tourism growth can be a “devil’s bargain” (Rothman, 1998). Zuo and Huang (2018) warn that the growth is difficult to maintain as they have observed an inverted-U-shaped relationship between tourism specialization and economic growth, consistent, they note, with the predicted stages of the Tourism Area Life Cycle (Butler, 1980). Current results, however, do not support this concern, as those counties with the greatest reliance upon tourism, as reflected in Table 1, also enjoyed the highest rate of NAICS-721 employment-growth.

A return to the literature regarding volatility similarly provides mixed recommendations. Casares (2013) suggested the larger an industry’s size, the lower its volatility. Contrarily, Hammond and Thompson (2004: 518) indicate that a “touchstone of regional policymaking has been the idea that regional specialization…will tend to create more variability in aggregate regional employment.” While their comment was directly related to economies with strong reliance upon a dominant industry, such as the coal mining region of West Virginia, it suggests that overdevelopment of tourism could be expected to result in high labor volatility, clearly a negative for the community. Supporting the latter argument, it is noteworthy that this research determined those counties with the highest tourism reliance also experienced the highest employment-volatility.

Another issue to consider was raised by Timo (1993: 37), who estimated that 65% of “tourism industry employment is less than full-time and predominantly casual.” This adds to volatility as eliminating part-time workers is generally easier than laying off one’s full-time employees (certainly observed during the COVID-19 experience). However, as with most issues, this too has multiple perspectives, for while there are clear advantages to having an economy strongly based on full-time employment, tourism provides many part-time opportunities for those unable to maintain a full-time position, often providing employment to workers who would otherwise be unemployed.

As a concluding comment, Berry and Blackwell (2005) specifically noted: “there may not be any such thing as an ‘unacceptable’ new job.” However, when allocating scarce resources within a community, government officials should consider both the positives of growth and the negatives of volatility when selecting among competing economic development alternatives. Such will be highly evident as post-COVID-19 stimulus decisions must be made and tourism officials seek to persuade governments that their industry is worthy of investment.

These findings are but a starting point. Communities must look at their own employment history and current dynamics to justify appropriate investment strategies. It is hoped that this research encourages such exploration and that their findings are of value to officials as they craft appropriate and effective economic tourism developmental plans for their community.

Before ending, there are several limitations to this study that should be considered. First, one must accept the fundamental assumption herein that hospitality employment serves as an effective proxy for tourism employment. While comfortable with the approach, future researchers are invited to identify alternative methods as they seek to confirm or extend the current work. In addition, the 5-year test period studied did not include a recessionary period and certainly did not envision the impacts of the COVID-19 pandemic. It would be of value to have this study repeated during a period of economic challenge, specifically as we recover from the current malaise, to determine if the findings are dependent on a generally positive economic environment or are consistent through both healthy and challenging economic times. Further, the measure of volatility could also be further considered. The data used reflected annual numbers. With tourism’s inherent seasonality, greater than for the general economy, additional volatility would have been evident had this work been based on monthly data. Finally, the research was based solely on US employment data. Replication elsewhere would be valuable.