Tourist displacement in two South African sport mega-events

Abstract

Crowding-out (or displacement) of non-event visitors has received little attention in the literature on the impact of sports events, largely because it cannot be measured accurately. This paper discusses such effects in conceptual terms and reports the results of an analysis of data on tourist arrivals in South Africa aimed at estimating the displacement effects of two sports events held in 2009: the Indian Premier League cricket tournament and the British and Irish Lions rugby tour. Using monthly tourist arrivals in South Africa from specific countries, we find that some tourists from countries not participating in these events were displaced; the much stronger effect, however, was that tourists from the participating countries re-arranged their visits to coincide with an event. While confirming the inherent difficulty of measuring crowding-out effects, this paper shows that characteristics of events can sometimes be exploited to obtain useful information on displacement from readily available data.

Keywords:

• sport tourism
• travel services
• crowding-out
• sport event
• mega-event
• cricket
• rugby union

JEL codes:

• L83

1. Introduction

The advent of professionalism in sport and the increasingly global reach of the broadcasting media have boosted the popularity of sport mega-events and countries compete fiercely to host them (Horne & Manzenreiter, 2006). The resources invested in preparing bids and the extensive media and popular interest in the selection process suggest it is widely believed that large benefits, both tangible and intangible, accrue to the hosts of such events.

The actual size of these benefits, though, remains unclear. One of the most notable features of economic impact assessments of sport mega-events is that ex ante estimates often wildly exaggerate the benefits to host cities and countries. In a review of economic impact studies, Matheson (2006:14) points out that the actual net benefits of sport mega-events sometimes amount to only one tenth of the ex ante estimates. Such discrepancies often stem from failure on the part of analysts to incorporate three sets of effects in ex ante impact studies (Matheson & Baade, 2004:1090–91; Matheson, 2006:8–13):

•	Substitution effects. Substitution effects occur when residents of the region where a sports event takes place, and foreigners who would have visited the region even if the event had not been staged there, spend money at the event instead of elsewhere in the local economy. Such expenditure switching does not increase aggregate spending in the local economy and therefore should not be regarded as a net benefit from hosting the event.
•	Crowding-out effects. Mega-events give rise to various effects that could prevent or dissuade potential tourists from visiting the area while they take place. Some contributors to the literature (e.g. Seaman, 2007) have distinguished between two types of crowding-out (also known as displacement): supply-side crowding-out, which occurs when capacity constraints related to the tourist infrastructure of host cities or regions (such as transport facilities, accommodation services and other amenities) restrict the number of non-event tourists while the event takes place, and demand-side crowding-out, which occurs when event-related price increases, safety and security risks or a dislike of various forms of congestion dissuade non-event tourists from visiting. Failure to consider such crowding-out effects would result in inflated estimates of the positive impact of sports events on tourist and spending flows.
•	Income and spending leakages. Almost all ex ante impact assessments correctly provide for multiplier effects; that is, additional rounds of expenditure induced by the direct spending associated with mega-events. The chosen values of the multipliers, however, often do not account properly for leakages from the circular flows of income and expenditure and therefore exaggerate the impact of events. Common examples of such leakages include payments for imported materials and remittance of wages and profits by foreign workers and hotel chains.

This paper examines the displacement (or crowding-out) effects of mega-events. It discusses such effects in conceptual terms and reports the results of an analysis of data on tourist arrivals in South Africa aimed at estimating the displacement effects of two sports events held in 2009, the Indian Premier League cricket tournament and the British and Irish Lions rugby tour. Crowding-out effects have received relatively little attention in the literature on the impact of sports events, largely because they are extremely difficult to measure (Fourie & Santana-Gallego, 2011). Awareness of crowding-out effects, however, is by no means a recent development. In a paper published more than 15 years ago, Crompton (1995) listed the neglect of displacement effects as one of 11 common shortcomings of economic impact analyses of sports events.

The issue of displacement has been highlighted most often in the context of the two major mega-events, the FIFA World Cup (Szymanski, 2002; Baade & Matheson, 2004; Allmers & Maennig, 2009) and the Summer Olympic Games (Baade & Matheson, 2003; Preuss, 2004, 2007b; Rose & Spiegel, 2011). However, studies of other sports events – including the 1995 World Athletics Championships in Göteborg in Sweden (Hultkrantz, 1998), the 1999 Rugby World Cup in Wales (Jones, 2001) and the 2009 US Open Golf Championship for Women in the LeHigh Valley (Scott & Turco, 2009) – have also referred to the phenomenon, as have studies of non-sports tourism (e.g. Brännäs & Nordström, 2006; Bresson & Logossah, 2011). This paper adds to this growing literature, by considering a natural experiment from two less well-known sport mega-events: the 2009 Indian Premier League and the 2009 British and Irish Lions Tour, both hosted by South Africa. The paper does not fully overcome the inherent difficulty of measuring crowding-out effects, but contributes to the literature on the topic by highlighting how characteristics of events (such as aspects of their scheduling) could be exploited to obtain useful information from readily available data.

2. The meaning and estimation of displacement

The term ‘crowding-out’ can assume various meanings in the context of sports mega-events. Some meanings relate more to the shorter-term economic impact of events, for example displacement of foreign visitors and displacement of expenditures by potential visitors and residents of host regions. Other meanings, for example displacement of urban space (crowding-out by tourism infrastructure of public facilities required by residents) and displacement of government priorities (for example, prioritisation of the safety and security and other needs of foreigners over those of the local population) pertain to the longer-term legacies of events.1 This study focuses on the displacement of foreign tourists during events; legacy aspects such as the effect of events on the provision of physical infrastructure and the policy priorities of the governments of host regions fall outside its scope. The remainder of this section elaborates further on the concept of displacement of foreign tourists and explains how it may be measured.

Assume that Region A hosts a sport mega-event in time period t, and let T _t ^A denote the total number of visitors to this region while the event takes place. This number is made up of those who visit Region A for the purpose of attending the event and visitors with other motives. If these two groups are denoted E ^A _t and N ^A _t , respectively, the total number of visitors can be written as follows:

The actual number of ‘non-event-linked’ visitors to Region A (N ^A _t ) is the residual obtained when subtracting two groups of people from all potential foreign visitors to the region during period t (P ^A _t ): those who had never intended to visit Region A during the event (U ^A _t ), and those who had planned to do so but were displaced as a result of its supply-side or demand-side crowding-out effects (D ^A _t ):

Equation 2 highlights the first major constraint to accurate measurement of crowding-out effects: displaced potential visitors cannot be identified, and surveys therefore cannot be used to obtain information about their numbers and what they had planned to spend in the host region. As Matheson and Baade (2004:1090) put it: ‘A fundamental shortcoming of economic impact studies pertains, therefore, not to information on spending for those who are included in a direct expenditure survey, but rather to the lack of information on the spending behaviour for those who are not.’ This reality severely circumscribes the scope for using bottom-up analysis based on micro-data to estimate the extent of visitor displacement. Bottom-up analysis is generally superior to top-down analysis for analysing the impacts and legacies of sports events, because the macro-data used for top-down analyses seldom allow researchers to distinguish between event-related and non-event-related changes in tourist numbers and consumption expenditure (cf. Preuss, 2007a).

Accurate estimation is complicated further by the existence of two types of displacement with very different effects on the impact of sport mega-events. Potential visitors who wish to avoid exposure to congestion effects could either cancel their planned visits altogether or reschedule them to periods before or after the event. Preuss (2005:288–9) labels the displaced the ‘avoiders’ and distinguishes between ‘cancellers’ and ‘pre/post switchers’. If these groups are denoted C ^A _t and S ^A _t , respectively, crowded-out potential visitors can be written as:

Although the members of both these groups of potential visitors change their plans so as to avoid Region A for the duration of the event, only the members of group C ^A _t are fully displaced in the sense that their planned visits to the region no longer materialise. The members of group S ^A _t are displaced temporarily, but the rescheduling of their plans does not deprive Region A of the benefits of having them as visitors. A similar distinction applies in the case of event-linked tourists to Region A (E ^A _t ). This group is made up of those who arranged visits with the sole purpose of attending the event, and those who would have visited the region during some other period, but rescheduled their visits to coincide with the staging of the event. Preuss (2005:287–8) labels these groups the ‘event visitors’ and the ‘time-switchers’, respectively. Only the benefits resulting from the visits of the ‘event visitors’ should be included in calculations of the net benefits to Region A of hosting the event. Further discussion of this aspect of the assessment of the impact of large sports events falls outside the scope of this paper.

Hence, C ^A _t should be the relevant aggregate in most efforts to quantify displacement effects. Top-down analysis is not well suited for this aspect of analysing crowding-out either, because aggregate data on tourist flows are likely to obscure the distinction between displacement proper and time-switching of visits.

It should also be noted that top-down analyses of visitor numbers cannot distinguish between event-linked and other tourists. Hence, such analyses implicitly assume that visits by event-linked and other tourists are perfect substitutes. This assumption is dubious: the lengths of stays and expenditure patterns of sports tourists often differ markedly from those of other travellers (see for example Turco et al., 2003).

Clearly, aggregate data on tourist flows are far from ideal for estimating the extent of crowding-out effects. The impossibility of identifying and surveying displaced tourists, however, dictates the use of top-down methods based on such numbers. The usual method is to analyse divergences between the actual numbers of visitors during events and estimates of the number that would have visited the region if the events had not taken place. Total visitor numbers and visitor numbers from specific countries or regions could be analysed in this manner. Such analyses can detect crowding-out when events did not boost visitor numbers (that is, when the actual numbers of visitors were smaller than or similar to ‘without-event’ counterfactuals), especially when the availability of country-level data on visitor numbers permits more disaggregated analysis. In South Korea, for example, the total number of foreign visitors during the 2002 FIFA World Cup was almost exactly the same as in the same period of the previous year, and data on arrivals indicated that the large inflow of European tourists during the tournament displaced similar numbers of Japanese business travellers and holidaymakers (Matheson & Baade, 2004:1090). Of course, the far more common outcome is that actual visitor numbers during events exceed counterfactual estimates. In such cases, top-down analysis has limited value for identifying and quantifying crowding-out effects, although otherwise unexplained decreases in the numbers of visitors from specific countries during events could indicate displacement. In the same way, such decreases in visitor numbers from specific countries during mega-events and increases in periods immediately before and after these events could denote the presence of temporary displacement.

3. Method and data

As was indicated earlier, this paper provides rough estimates of the displacement effects of the 2009 Indian Premier League cricket tournament and the British and Irish Lions rugby tour to South Africa in 2009.2 This section discusses the methods and data used in the analysis.

The Indian Premier League (IPL), which was staged for the first time in 2008 to capitalise on the growing popularity of the shortest version of cricket (20 overs a side), is contested by eight franchised teams consisting of the best Indian cricketers as well as star players from other countries. At the end of March 2009, a mere three weeks before the second season was due to begin in India, the entire tournament was moved to South Africa because of security concerns linked to the general elections in India. The tournament took place from 18 April to 24 May 2009 and consisted of 59 matches. The British and Irish Lions is a rugby union team consisting of top players from England, Ireland, Scotland and Wales. The Lions squads tour one of the three major rugby-playing countries in the southern hemisphere (Australia, New Zealand and South Africa) every four years on a rotational basis. The 2009 Lions tour to South Africa took place from 30 May to 4 July and consisted of a three-match Test series against the South African Springboks and seven other matches against South African selections.

Although not comparable in size to the Olympic Games or the FIFA Soccer World Cups, both the IPL and the Lions tour were major sporting occasions that should have had significant impacts on the South African economy. The inaugural IPL held in India in 2008 drew 3 422 000 live spectators – an average of 58 000 for each of the 59 matches (Wikipedia, 2009). The number of tourists visiting South Africa would have been boosted markedly if only a small fraction of these fans had travelled to support their teams during the 2009 competition. An impact study commissioned by SA Rugby reported that the Lions tour attracted some 37 000 visitors from Britain and Ireland (SAPA, 2009). The net economic benefits of the two events should have been raised further by (a) the ready availability of suitable facilities (neither required large capital outlays on the construction or expansion of stadiums, which are often major expenses associated with the hosting of such events) and (b) their scheduling outside the peak tourism season in South Africa. Matheson uses the following example to underscore (a):

The local government of Montreal built multiple new facilities for the 1976 Summer Olympics, including the grandiose Olympic stadium, and wound up with debts totalling $1.2 billion. These debts were not paid off until 30 years after the Games. In contrast, the 1984 Los Angeles Olympic Committee exclusively used existing sports venues around the city, spent less than $1 billion in total to put on the Games, and ended up with a profit of over $200 million. (2006:19)

With relevance to (b), Higham (2005) points out that displacement tends to be greater during peak tourist seasons (when tourism infrastructure is more likely to be subject to capacity constraints) than during off-seasons.

3.1 Method

This paper uses the counterfactual-based method outlined above to estimate displacement. Preuss (2007a:215–17) discusses two methods for obtaining counterfactual estimates of tourist numbers. The first uses data from other cities in the same economy with roughly similar sizes and economic structures to construct reference cases, while the second uses model-based forecasts. The reference-case method could not be used in this study, because both the IPL and the Lions tour involved matches in several South African cities of varying sizes and structures. Hence, the counterfactual estimates of the number of foreign tourists to South Africa in 2009 used in the analysis below are projections based on econometric modelling of all arrivals from overseas countries as well as arrivals from selected countries. These countries are China, France, Germany, India, the Netherlands, Spain, the UK and the USA. Detailed analysis of arrivals from India and the UK was required because the vast majority of tourists who attended the IPL and Lions tour matches hailed from these countries.

In contrast to the Olympic Games and FIFA World Cups, the IPL and Lions tours are not global events. Fully 83% of the international visitors who attended matches of the British and Irish Lions in New Zealand in 2005 were from Britain and Ireland, while some 10% were from Australia and the remaining 7% were from other countries (Covec Ltd, 2005:11). It is unlikely that many Australian rugby fans would have undertaken the much longer journey to South Africa for the 2009 Lions tour. Similar data are not available for the IPL, because the competition had not been staged outside India before. To the best of our knowledge, however, no one has claimed that the 2008 IPL attracted large numbers of visitors from outside the subcontinent to India.

In addition, the UK is one of South Africa's five primary overseas markets as far as inbound tourists are concerned, along with France, Germany, the Netherlands and the USA. China has become an increasingly important market for the South African tourism industry in recent years. Finally, Spain was analysed separately because the number of Spanish visitors to South Africa was boosted in 2009 by the country's participation in the FIFA Confederations Cup football tournament, which partly overlapped with the British and Irish Lions tour. The Confederations Cup, which involved the national teams of eight countries, took place in South Africa from 14 to 28 June 2009 as a prelude to the 2010 FIFA World Cup. Literature on models for forecasting tourist arrivals suggested reliable control variables for inclusion in the model (see below).

As was indicated above, estimation of displacement effects with aggregate data on visitor numbers is a difficult undertaking which yields rough orders of magnitude rather than precise numbers. Some features of the two events facilitated estimation. The low probability that visitor numbers from any countries apart from India and Britain and Ireland were boosted significantly by the IPL and the Lions tour, respectively, suggests that decreases in visitor numbers from non-participating countries during these events could be regarded as upper-bound estimates of event-related crowding-out (provided, of course, that the counterfactual estimates are reliable). Furthermore, otherwise unexplained increases in the numbers of visitors from India and the UK in the months prior to and after the IPL and the Lions tour could be interpreted as upper-bound estimates of temporary (i.e. time-switching) displacement from these countries. In the case of the hastily arranged IPL, displacement of visitors from India and elsewhere was probably small and restricted to the months after the tournament. By contrast, individuals and groups who had planned to visit South Africa in June 2009 had ample time to reschedule their planned visits to periods before or after the Lions tour. Hence, temporary displacement related to the Lions tour was probably more extensive and more dispersed than that resulting from the IPL. Unfortunately, the method used in this paper was inadequate for detecting and quantifying permanent displacement of visitors from participating countries during the event periods (i.e. visitors from India in April and May 2009 and visitors from the UK in June 2009).

3.2 The model

This paper uses a popular technique for modelling tourist demand. It consists of a log-linear single equation model, with independent and dependent variables expressed in logarithmic form. The independent variables in the model are the income of the country of origin of tourists (gross domestic product or GDP), relative prices in the destination (proxied by the real exchange rate) and the difference in transport costs between the two destinations concerned (proxied by oil prices). These three variables are often used in studies on the determinants of tourist flows. Sinclair (1998) and Saayman & Saayman (2008) are two examples of such studies. Hence, the basic specification of the model is:

Equation 4 expresses tourist arrivals (y _t ) as a function of a vector of explanatory variables (x _t ), a vector of dummy variables (z _t ) and a homoscedastic error term with a zero mean (ϵ _t ).

The vector of explanatory variables consists of the natural log of GDP (lgdp _t ), the natural log of the real exchange rate (lrex _t ) and the natural log of the price of Brent crude oil (loil _t ):

The vector of dummy variables consists of event-specific dummies (event _t ), a time indicator variable to account for long-run idiosyncratic shocks (time _t ) and a month indicator which is included to account for seasonal factors (month _t ):

Table 1 shows the dates during which the events took place and the corresponding months in which the event dummies took the value of 1.

Table 1: Event dummies

Event	Dates	Dummy variable month
Indian Premier League	18 April–24 May 2009	May
British and Irish rugby tour/Confederations Cup	25 May–24 June 2009/14 June–28 June 2009	June

Tourists base their decisions to travel on past and present costs and income levels; resolving such autocorrelation in the dependent variable requires a dynamic model that includes lagged variables. The model therefore includes lagged variables of the dependent as well as the independent variables. A dynamic specification of the model which includes the lags (j) possible for the period (p) and where the vector x _t now also includes a lagged dependent variable y _t-j is:

Separate models were run for each event to forestall the possibility of cointegration of the event indicator variables. To avert serial correlation, a vector auto-regression estimate was run to provide a basis for determining the lag lengths. The presence of first-order serial correlation was identified by the Durbin-Watson statistic. The results of the vector auto-regression suggested the lag length of the variables for different periods, with each period taken as being one month long. The results of these diagnostic tests are available from the authors on request.

The best specification for each country was obtained by means of repeated estimation of the models, informed by the values of the Akaike information criterion and the Schwarz criterion.

3.3 The data

3.3.1 Data on tourist arrivals

This paper uses the number of inbound tourists to South Africa as the dependent variable to measure the impact of hosting large sports events on tourist arrivals. In principle, the extent of inbound tourism could also be measured in terms of the total expenditure by or receipts received from the tourists concerned. Saayman andSaayman, however, found that the available data on tourist expenditures and receipts in South Africa are inconsistent (2008:84). Numbers of inbound tourists are readily available for arrivals from all countries, being published by Statistics South Africa in a monthly statistical release entitled Tourism and Migration (P0351).3 The data series used here runs from January 1983 to July 2009. Statistics South Africa defines ‘foreign visitors’ as all visitors who are neither citizens nor permanent residents of South Africa.

Figure 1 shows that arrivals of foreign travellers in South Africa increased gradually from 704 444 in 1965 to 930 393 in 1989, but rose much more rapidly thereafter to reach 10 098 306 in 2009. A distinctive feature of South African tourist arrivals data is the predominance of residents of neighbouring countries, especially Botswana, Lesotho, Mozambique, Swaziland and Zimbabwe. In 2009, for example, fully 75% of all foreign visitors to South Africa hailed from the SADC countries (StatsSA, 2010:12). Saayman and Saayman point out that many of these SADC visitors classified as holidaymakers are actually migrant workers (2008:85). To prevent distortions arising from misclassification of such visitors, this paper focuses only on overseas visitors to South Africa and excludes all visitors from African countries from the analysis. Reflecting the impact of the global economic crisis, the number of overseas visitors to South Africa decreased by 9.2% from 2 514 771 in 2008 to 2 282 371 in 2009.

Figure 1: Arrivals of foreign travellers in South Africa (1983–2009)

The period during which the IPL and the Lions tour took place (mid-April to the end of June) has traditionally been an off-peak period for the tourist industry in South Africa. From 1983 to 2009, for example, the months of April and especially May and June attracted fewer foreign visitors, on average, than the other nine months of the year (cf. Figure 2). This aspect of the scheduling of the two events probably reduced supply-side crowding-out during the events.

Figure 2: Average monthly arrivals of overseas travellers in South Africa (1983–2009)

3.3.2 Data on income of the countries of origin

As was indicated above, the GDP was used in the models as a measure of the income of the countries of origin of foreign tourists to South Africa. Quarterly data were sourced from the IMF's World Economic Outlook and were interpolated linearly to obtain monthly figures.

The GDPs of all the countries included in the study except India were expressed in terms of their national currencies. Furthermore, the data were converted from nominal to real series using GDP deflators for each country with 2005 figures equal to 100. The deflators were obtained from the IMF's International Financial Statistics. In the case of India, the GDP in purchasing power parity terms was used. The Indian GDP was measured annually in billions of US$ in the relevant period.

3.3.3 Data on relative prices

The relationship between prices in the country of origin and the destination country is an important determinant of tourist flows. The potential tourist's currency is worth more in the destination country if its prices are lower (on average) than those of the country of origin. All else being equal, relatively low prices should therefore attract more tourists to a particular destination.

The rates of exchange (all expressed as rands per unit of foreign currency) between the South African rand and the following currencies were used in the models to proxy relative prices: the Chinese yuan, the Indian rupee, the US dollar (US$), the British pound (£) and – for France, Germany and the Netherlands – the European Union euro (€). Monthly averages of the rates of exchange between the rand and the US dollar, the rand and the British pound, and the rand and the euro were obtained from the South African Reserve Bank's monthly data releases. Yuan–dollar and rupee–dollar exchange rates obtained from the IMF's International Financial Statistics were converted to equivalent rand–yuan and rand–rupee exchange rates.

Following these conversions, nominal rand-denominated exchange rates were available for all the currencies. As the next step, real exchange rates were calculated to control for differences in the inflation rates of South Africa and the countries of origin. Such conversions were done using consumer price index data for South Africa and the countries of origins obtained from the IMF's International Financial Statistics. Data for Germany were available only for the post-unification period; that is, the period from January 1991 to June 2009.

3.3.4 Data on travel costs

The cost of transport to regions hosting large sports events is a major element of the total travel costs of tourists (including sports fans). This is especially true as far as mega-events in South Africa are concerned, because long-haul flights are the predominant mode of travel for visitors from primary overseas markets. As was indicated earlier, this study used monthly data on the dollar price of Brent crude oil as a proxy for travel costs to South Africa (such data have been published by the South African Reserve Bank since January 1981). This choice of a proxy was informed by research by Saayman and Saayman (2008), who reported that the dollar price of Brent crude oil correlated somewhat better with South African tourist arrivals than the price of jet fuel. Although the oil price has only an indirect effect on travel costs, the widely reported prices of Brent crude send signals which influence potential travellers' expectations of travel costs. In practice, the perceived costs of transport may well influence travel decisions as much as the actual prices of airline tickets.

4. Results

4.1 Evidence of ‘additional’ arrivals

Tables 2 and 3 present the event-dummy coefficients of the regression results by country of origin for several periods. There is clear evidence that the two events brought ‘additional’ tourists from those countries whose sportsmen participated in the event. Table 2 shows statistically significant increases of between 43 and 61% in the number of Indian tourists arriving in South Africa during the IPL. Similarly, Table 3 shows statistically significant increases of between 57 and 78% in arrivals of tourists from the UK during the British and Irish Lions tour. There is also evidence of an increase (between 19 and 65%) in arrivals from Spain during the month of the Confederations Cup; this relationship, however, is not consistently significant in statistical terms. There is also no statistically significant evidence of an increase in the number of tourists from the USA during the same period.

Table 2: Results of the Indian Premier League (IPL) estimates

Country	Change of arrivals (%)	Size of µ	Std error	T-statistic
China	–3.62	–0.036162	0.366574	–0.098648
1999/01–2009/06	–6.26	–0.062556	0.262262	–0.238526
2005/01–2009/06	–8.67	–0.086657	0.193262	–0.448391
France	–7.25	–0.0725	0.2098	–0.3454
1989/01–2009/06	–2.22	–0.022212	0.210515	–0.105514
1999/01–2009/06	1.32	0.013178	0.109806	0.120014
2005/01–2009/06	5.62	0.056184	0.070558	0.796284
Germany	–17.54	–0.1754	0.1766	–0.9934
1999/01–2009/06	–17.44	–0.174361	0.102454	–1.701843
2005/01–2009/06	1.90	0.019008	0.092703	0.205037
India	59.68*	0.5968	0.2508	2.3798
1989/01–2009/06	43.39*	0.433857	0.200157	2.167582
1999/01–2009/06	55.55*	0.555468	0.154129	3.603923
2005/01–2009/06	60.85*	0.60854	0.121491	5.008941
Netherlands	6.11	0.0611	0.2681	0.2279
1989/01–2009/06	0.28	0.002814	0.271189	0.010375
1999/01–2009/06	–2.16	–0.021611	0.130276	–0.165884
2005/01–2009/06	12.00	0.120019	0.107401	1.117485
Spain	–18.47	–0.184687	0.254783	–0.724878
1989/01–2009/06	–3.26	–0.032611	0.250522	–0.130172
1999/01–2009/06	–10.30	–0.103036	0.199498	–0.516479
2005/01–2009/06	–00.02	–0.000287	0.207189	–0.001384
UK	–28.63	–0.2863	0.1877	–1.5248
1989/01–2009/06	–22.46	–0.224618	0.175015	–1.283425
1999/01–2009/06	–37.89*	–0.378902	0.096199	–3.938739
2005/01–2009/06	–43.51*	–0.43506	0.112043	–3.882966
USA	1.66	0.0166	0.1578	0.1051
1989/01–2009/06	2.03	0.020256	0.151236	0.133939
1999/01–2009/06	2.10	0.020987	0.083553	0.251182
2005/01–2009/06	5.03	0.050287	0.082667	0.608311
Notes: Asterisks indicate significance at the 5% level. Bold type for India indicates that teams from India participated in the IPL.

Table 3: Results of the Lions tour estimates

Country	Change of arrivals (%)	Size of µ	Std error	T-statistic
China	–25.23	–0.252315	0.362243	–0.696534
1999/01–2009/06	–6.54	–0.065404	0.263726	–0.248000
2005/01–2009/06	–14.80	–0.147976	0.205577	–0.719810
France	–22.18	–0.2218	0.2094	–1.0589
1989/01–2009/06	–14.16	–0.141605	0.21221	–0.667287
1999/01–2009/06	–3.75	–0.037537	0.112286	–0.334303
2005/01–2009/06	–7.39	–0.073874	0.083266	–0.887209
Germany	–0.97	–0.0097	0.1832	–0.0527
1999/01–2009/06	–1.92	–0.019231	0.110549	–0.173959
2005/01–2009/06	9.30	0.093033	0.103248	0.901066
India*	–54.30*	–0.5430	0.2531	–2.1453
1989/01–2009/06	–65.29*	–0.652864	0.199182	–3.277732
1999/01–2009/06	–46.42*	–0.464188	0.172603	–2.689343
2005/01–2009/06	–60.89*	–0.608885	0.193447	–3.147561
Netherlands	4.36	0.0436	0.2725	0.1600
1989/01–2009/06	1.15	0.011475	0.276991	0.041427
1999/01–2009/06	–4.15	–0.041522	0.134717	–0.308219
2005/01–2009/06	–5.29	–0.052905	0.124882	–0.42364
Spain	19.47	0.194661	0.254591	0.764603
1989/01–2009/06	34.16	0.341638	0.252877	1.351007
1999/01–2009/06	65.74*	0.657481	0.203271	3.234502
2005/01–2009/06	33.27	0.332754	0.238479	1.395319
UK*	57.25*	0.5725	0.1900	3.0128
1989/01–2009/06	78.01*	0.780078	0.177872	4.385601
1999/01–2009/06	69.96*	0.699583	0.096893	7.220193
2005/01–2009/06	67.68*	0.676756	0.096154	7.038221
USA	–7.26	–0.0726	0.1589	–0.4568
1989/01–2009/06	0.50	0.004983	0.152919	0.032583
1999/01–2009/06	–3.00	–0.029973	0.086001	–0.348521
2005/01–2009/06	–6.41	–0.064087	0.088943	–0.72054
Notes: Asterisks indicate significance at the 5% level. Bold type indicates that the Lions team is from the UK, and that the national football teams of Spain and the USA participated in the Confederations Cup tournament, which partly overlapped with the British and Irish Lions tour.

4.2 Evidence of displacement

The most surprising result is the extent of displacement. Arrivals of British tourists contracted significantly (by between 22 and 43%) in May 2009, the month in which the IPL took place in South Africa. The rather weak evidence of concurrent falls in the number of arrivals from other regions suggests that the displacement of UK tourists largely took the form of time switching: visits were shifted from May to June 2009 to coincide with the British and Irish Lions tour. The recorded year-on-year increase in the number of arrivals from Britain in June 2009 (when the Lions tour took place) should therefore be interpreted against the background of the drop in arrivals from the same country in the month before the event. Clearly, the 57% increase in arrivals from the UK in June in our base model looks less impressive if the 28% decrease in the previous month (May 2009) is taken into account.

The relatively stable arrival coefficients for the other countries in our IPL model suggest that the event had little impact on tourist behaviour in May 2009. This finding accords with our natural experiment prediction. However, while statistically insignificant, there seem to be some negative impacts on tourist arrivals from countries that did not participate in the Confederations Cup or the British and Irish Lions tour, especially China and France. The two other major tourist markets, Germany and the Netherlands, however, reveal no deviation from the norm. There is thus some (although weak) evidence that a mega-event (of the size of the IPL, Confederations Cup and British and Lions tour) displaces tourists from countries not participating in the event. The dominant effect measured here, though, is that of tourists from the participating country shifting their visits to coincide with an event.

5. Conclusions

Crowding-out (or displacement) of non-event visitors has received little attention in the literature on the impact of sports events, largely because it cannot be measured accurately. The phenomenon has been interpreted in many ways and this paper aimed primarily to shed more light on these interpretations, in particular with the aim of isolating the ‘detrimental’ types of displacement. The conceptual discussion also highlighted the pros and cons of the different types of measurement technique. Using a standard econometric approach and publicly available data, the paper reported the regression results of tourist arrivals in South Africa, with the aim of estimating the displacement effects of two sports events held in 2009: the Indian Premier League cricket tournament and the British and Irish Lions rugby tour.

The findings suggest that some tourists from countries that did not participate in these events were displaced; the much stronger effect, however, was that tourists from the participating countries re-arranged their visits to coincide with an event. While confirming the inherent difficulty of measuring crowding-out effects, the paper showed that characteristics of events can sometimes be exploited to obtain useful information on displacement from readily available data.

Notes

¹For such a discussion, see Preuss (2007a:212–13).

²For an assessment of earlier South African mega-sports events, see Fourie & Spronk (2011).

³It was assumed that the reported numbers of incoming arrivals during the events did not include any transfer passengers.