How Does the Weather Influence Travel Activity? Evidence from Michigan

How Does the Weather Influence Travel Activity? Evidence from Michigan

Submitted for review for the 2005 Annual TTRA Conference

by

Charles Shih and Sarah Nicholls

Department of Community, Agriculture, Recreation & Resource Studies

Michigan State University

131 Natural Resources Building

East Lansing, MI 48824-1222

Phone: (517) 353 5190

Email

Charles Shih:

Sarah Nicholls:

Submitted for review as an oral presentation only (option a)

Session topic “travel research methods, approaches or uses”

How Does the Weather Influence Travel Activity? Evidence from Michigan

Given the heavy dependence of a large number of tourism activities on the appropriate weather conditions, either to make them possible, e.g., adequate snow for skiing, or, at least, enjoyable, e.g., warm enough conditions for beach activities, the lack of published research on the relationships between participation and the weather is surprising. In light of increasing evidence regarding climate change, most notably the global trends towards rising temperatures and sea levels (IPCC, 2001), this lack of knowledge should be of special concern to tourism providers worldwide. Despite the call by Butler (2001) for the industry to pay much closer attention to this issue, climate change remains largely absent from the tourism literature.

Rising temperatures have the potential to negatively impact both winter and summer destinations, by reducing the length of seasons for winter activities such as skiing while leaving some summer destinations such as those in southern Europe simply too hot for human comfort. The impacts of projected climate change will vary spatially, however, and many regions which currently experience somewhat marginal weather may in the future experience more pleasant conditions, thereby increasing their attractiveness as seasonal or year-round tourism destinations (Agnew and Viner, 2001). The outlook for the tourism industry is, therefore, mixed, and further research is crucial so as to best equip the industry for successful adaptation to future conditions.

Analysis of the likely impacts of projected climate change on tourism activities and patterns first requires understanding of current relationships between participation and the weather, and it is this dynamic that forms the basis of this study. Specifically, the purpose of the paper is to ascertain the major influences and, hence, the relative significance of weather variations, on tourism traffic in the state of Michigan. The somewhat generic measure of “tourism traffic,” rather than levels of participation in individual activities such as skiing or camping, was chosen to represent the overall level of tourism activity in the state, and was considered appropriate since the majority of Michigan travelers use cars and other recreational vehicles as their primary mode of transportation (as noted by, e.g., Rink, 2004).

Methods

Daily highway traffic counts were obtained from the Michigan Department of Transportation for a 10-year period (1991-2000) for 18 “class counters.” Such counters record the number of axles on each vehicle, thereby allowing the removal of large vehicles (four or more axles), likely to be traveling for commercial rather than tourism purposes, from the dataset. Whereas previous research has focused mainly on the influence of income and prices on aggregated consumption data such as annual visits (e.g., Crouch, 1994; Lim, 1999), the use of daily data enables the investigation of much shorter-term variations in the explanatory variables.

The following independent variables were available for inclusion in the analyses: daily maximum temperature, daily precipitation, economic conditions (as suggested by the monthly Consumer Confidence Index published by The Conference Board), average weekly gas prices (as recorded by the American Automobile Association), time of week (specifically, Monday-Thursday, Friday or Sunday, or Saturday), public holidays (plus the traditional travel days immediately before and after such holidays), and year (to account for time series issues). Weather data were acquired from the recording station closest to each traffic counter.

Separate regressions were run for each season, namely winter (Dec-Feb), spring (Mar-May), summer (Jun-Aug) and fall (Sep-Nov). Experimentation with various forms indicated that a logarithmic transformation of the dependent variable produced the most useful models. The regressions each followed the general form: log (traffic) = f (temperature, precipitation, consumer confidence, gas price, day of week, public holidays, year).

Findings

Model performance (as measured by R2) varied substantially with the type of route on which the traffic counter was located, classified by the authors as either urban, rural, or tourist. The lowest levels of performance (R2 below 0.4) were discovered for the four counters on rural roads, routes along which the majority of traffic is likely to be local, e.g., M-95 in the Upper Peninsula. In urban areas, e.g., along M-10 in Detroit, performance was somewhat improved, ranging from 0.4 to 0.6 across the six counters. However, much of this traffic is also likely to be local, business, and/or commuter. The highest levels of performance (R2 above 0.6) were recorded for the seven counters on roads likely to receive rather large proportions of tourist traffic, e.g., US-127 in Clare, one of the most heavily utilized routes between the lower half of Michigan and the northern parts. Since the focus of the study was primarily on building models of tourism-related movement, the following observations refer only to the seven counters located along corridors dominated by tourist traffic. Table 1 demonstrates a sample set of results.

Table 1. Sample Results for Class Counter on US-127 at Clare, Michigan

Season / Spring / Summer / Fall / Winter
Independent Variables / β / Sig. / β / Sig. / β / Sig. / B / Sig.
Constant / -84.46 / 0.00 / -71.89 / 0.00 / -85.27 / 0.00 / -80.29 / 0.00
Maximum temperature / 0.02 / 0.00 / 0.01 / 0.00 / 0.02 / 0.00 / 0.00 / 0.45
T-T* (spline term) / - / - / -0.07 / 0.05 / - / - / - / -
Precipitation / 0.00 / 0.08 / 0.00 / 0.45 / 0.00 / 0.00 / -0.01 / 0.00
Consumer confidence / 0.00 / 0.20 / 0.00 / 0.09 / 0.00 / 0.03 / 0.00 / 0.35
Gas prices / 0.19 / 0.04 / -0.24 / 0.00 / -0.12 / 0.28 / -0.12 / 0.17
Fri/Sun (non-holiday) / 0.76 / 0.00 / 0.88 / 0.00 / 0.85 / 0.00 / 0.80 / 0.00
Saturday (non-holiday) / 0.27 / 0.00 / 0.40 / 0.00 / 0.29 / 0.00 / 0.26 / 0.00
Holiday weekend (begin/end) / 1.10 / 0.00 / 1.17 / 0.00 / 1.11 / 0.00 / 0.71 / 0.00
Holiday weekend (middle) / 0.66 / 0.00 / 0.62 / 0.00 / 0.55 / 0.00 / 0.44 / 0.00
Year / 0.05 / 0.00 / 0.04 / 0.00 / 0.05 / 0.00 / 0.04 / 0.00
Model R-square / 0.83 / 0.83 / 0.83 / 0.74

Time appeared to have the most substantial impact on tourist traffic across all locations and all four seasons. In almost all cases, the variables representing weekends (Friday-Sunday) and holidays were statistically significant at the 5% level or better. On non-holiday weekends, Fridays and Sundays exhibited substantially increased levels of traffic, while the increase on Saturdays was less dramatic though still statistically greater than during the week. As expected, holidays also tended to be associated with substantial increases in traffic, with the greatest rises on the traditional travel days immediately preceding and following these days. The year variable, representing an annual time trend, consistently indicated that traffic along the routes from which data were available steadily increased by about 4% per annum over the time span studied.

Of the two weather variables available, maximum temperature appeared to have the most substantial impact on tourist traffic. Temperature had a highly significant positive impact on traffic at all counter locations in spring, summer, and fall, but not in winter. In general, traffic was found to increase by 1 to 2% with a 1ºC rise in temperature. Further analysis utilizing a spline term, however, a technique that enables identification of shifts in a coefficient’s magnitude at some particular level (Greene, 2002), suggested that a threshold temperature does indeed exist, with summer traffic starting to decrease when temperature rises above 32ºC or 90ºF. Precipitation appeared to have a lesser impact on tourist traffic than maximum temperature; its effect was found to be significant in the fall and winter seasons only, for five of the seven counters.

Of special interest given current economic and political circumstances, gas prices were found to have little to no impact on tourist traffic. Similarly, the variable representing consumer confidence rarely reached statistical significance, again suggesting that economic factors may have less influence on tourist traffic than other factors such as time and temperature.

Application of Results

The results suggest that the weather does indeed have a statistically significant impact on tourism traffic in Michigan. Since previous analyses of these relationships have been rare, results such as these, which more clearly articulate their magnitudes and directions, should be of great utility to a multitude of tourism providers, for both their short and long term planning and management decisions. This is especially the case for those offering traditional winter and summer activities, such as skiing, camping and golfing, which are particularly weather-dependent. The models also demonstrate the relative impact of temporal and economic factors, with the former appearing to exert significantly more influence on tourism travel in Michigan than the latter, a finding that appears to contradict some surveys as well as popular opinion regarding the deleterious impact of higher gas prices on automobile traffic (e.g., Jones, 2000).

Further, the models constructed would form the basis of any analysis of how future climate variability and change might impact the outdoor recreation and tourism sectors. For example, examination of traffic levels at one, two and three standard deviations above the historical mean maximum temperature would reveal how travel activity might respond to the warmer conditions suggested by most climate scientists. Similarly, potential deviations in other weather variables, as well as in other economic, social or demographic conditions, could be incorporated, providing the industry with a wide range of future scenarios to consider and plan for. Acquisition of daily participation data for specific activities or from individual sites would allow more targeted analyses than the somewhat generic, traffic-based models presented here.

Conclusions

Given the highly intuitive nature of the relationship between weather conditions and tourism activity, the lack of research on this topic is surprising. While historical weather and future climate data are plentiful, perhaps the greatest challenges facing tourism-related research in this area are the raising of awareness of the issue, and the acquisition of historical participation data upon which to conduct analyses of past and future trends. Once such data have been acquired, our research suggests that it is indeed possible to construct rigorous models that can then be adapted into highly meaningful decision support tools for the industry. Future research will focus on integrating the models of tourist traffic with various climate change scenarios, as well as on the construction of participation models for specific sectors of the Michigan tourism industry such as skiing, camping and golf.

References

Agnew, M., & Viner, D. (2001). Potential impacts of climate change on international tourism.

Tourism and Hospitality Research, 3(1), 37-61.

Butler, R. (2001). Editorial. Tourism and Hospitality Research, 3(1), 5-6.

Crouch, G. I. (1994). The study of international tourism demand: A review of findings. Journal

of Travel Research, 33(1), 12-23.

Greene, W. H. (2002). Econometric analysis. (6th. ed). New Jersey: Prentice Hall.

IPCC. (2001).Climate Change 2001: The ScientificBasis. Contribution of Working Group I to

the Third Assessment Report of the Intergovernmental Panel on Climate Change [Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K., and Johnson C.A. (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

Jones, J.M. (2000). High gas prices affecting many Americans' driving and travel plans.

Retrieved January 13, 2005, from

Lim, C. (1999). A meta-analytic review of international tourism demand. Journal of Travel

Research, 37(3), 273-284.

Rink, J. (2004). 1.1 Million to travel during Labor Day, says AAA Michigan. Retrieved January

14, 2005, from articleID=1433&view=archive

1