Economic incentives for tourists: Evaluating a regional policy measure

Abstract

This Note examines the effectiveness of a regional policy centered around offering discounted vouchers to tourists. It specifically studies the case of Sicily, which, unlike other Italian regions, implemented this policy between 2021 and 2023 in response to the aftermath of the COVID-19 shock. Difference-in-differences estimates reveal that this intervention has failed to significantly affect tourism.

Keywords

diff-in-diff model policy evaluation tourism voucher

Introduction

This Note contributes to the growing body of literature exploring recent public interventions in various countries aimed at supporting tourism through the distribution of complimentary vouchers, particularly following the COVID-19 lockdowns. Notable examples of such research include the studies by Toubes et al. (2021) on the Basque regions in Spain, Alba and Popescu (2023) on Romania, and Matsuura and Saito (2022) investigating Japan.¹ These studies emphasize the crucial role of holiday vouchers or price subsidies in reviving the tourism industry. Apart from Matsuura and Saito (2022), they predominantly focus on evaluating the used vouchers, overlooking the evident recovery of the tourism sector post-lockdowns. However, assessing the impact of subsidies introduced to stimulate post-COVID-19 tourism faces challenges, particularly in distinguishing the specific effects of the subsidy from the overall growth observed following the relaxation of restrictions.

This study examines the specific case of the Sicilian Region, which, unlike other regions in Italy, implemented a voucher system for tourists. Employing a difference-in-differences (“diff-in-diff”) approach—where the subsidy represents the treatment administered in certain destinations and not in others—allows us to evaluate the genuine impact of the voucher on the recovery of tourism.

Specifically, in Autumn 2020, the Sicilian region in Italy approved the “SeeSicily” program to boost tourism following the COVID-19 lockdown. This program provides a complimentary third night’s accommodation through a voucher for visitors booking a minimum two-night stay in an accommodation facility, such as a hotel or other lodging, within the region. Tourists can also choose one additional service like excursions, guided tours, diving experiences, or museum visits. The program includes a 50% refund on transportation tickets. This incentive extends to domestic and international tourists, regardless of their income, occupation, age, or other factors. To benefit, tourists must buy a package from a registered provider, such as a travel agent or tour operator, listed as part of an agreement between the Region and these agencies.

The incentives apply to stays from September 2021 onwards, through 2022 (excluding July and August, and until September 2023 (still excluding July and August). Clearly, the promotion aims to contribute to de-seasonalizing tourism, which typically peak during the sea&sun months, and to increase average stays.²

The initiative was funded with 75 million euros (approximately 37 million for accommodations, 20 million for guided tours, excursions, and museum visits, 13 million for transportation, and 5 million for advertising and promotion). At present, official figures regarding the number of tourists benefiting from this promotion are unavailable, sparking political debate over the scale and impact of public spending.

Nevertheless, official data on tourist arrivals and stays—sourced from ISTAT, the Italian Statistics Office—are already accessible, even for 2022. These data can provide pieces of evidence, allowing us to assess the effectiveness of this public policy measure in stimulating tourism. To achieve this objective, a diff-in-diff model is employed, using data of tourist flows to municipalities in Sicily before and after the campaign’s implementation (the treatment group), compared with the data from municipalities in regions where the promotion was not introduced. We use here the municipalities of Sardinia as the control sample, where such an incentive has not been introduced; however, the evaluation exercise can be replicated using other regions as controls. We present the results when comparing Sicily and Sardinia, for the obvious reason that these are the two Italian island regions, sharing common challenges related to insularity and their reliance on sea&sun tourism.

We consider both arrivals and overnight stays, both domestic and international tourists. The year in which the policy takes effect (i.e., the treatment is operative) is 2022. This year’s data are compared with that of 2019, which serves as the pre-treatment period. The results are striking: in no case does the policy measure appear to be effective according to diff-in-diff models. This holds true for arrivals and stays, for both domestic and international tourism.

Model and data

We estimate a basic diff-in-diff model,³ with the variable under consideration in level:

y_{i, t} = α + β R_{i} + γ T_{t} + δ (R_{i} ∙ T_{t}) + e_{i, t}

(1)

Variable $y_{i, t}$ denotes the tourism flow variable in municipality i at time t, with t = 2019 or 2022 (the pre-treatment and post-treatment year, respectively).⁴ Variable $R_{i}$ is the dummy variable associated with the location in region Sicily where the treatment has been applied ( $R_{i} = 0$ for municipalities in Sardinia and $R_{i} = 1$ for municipalities in Sicily). Variable $T_{t}$ is the dummy associated with the time when the treatment is in effect ( $T_{t}$ = 1 for t = 2022, and 0 otherwise). As usual, coefficient $δ$ captures the specific effect of the treatment. The sample comprises all municipalities in Sicily and Sardinia for which data on tourism flows are available.⁵

We also resort to the dataset consisting of joint-repeated data only, that is, retaining in the sample only those municipalities for which data are available both before and after the treatment; in this scenario, the regression equation is:

∆ y_{i} = α + δ R_{i} + u_{i} .

(2)

The number of observations (i.e., municipalities) available for estimating models (1) and (2) is reported in Table 1. Note that a few municipalities are present in only 1 year under consideration and are therefore omitted from the estimation using joint-repeated data. All data used in this study are freely downloadable from the ISTAT website.

Table 1.

Number of municipalities in the sample.

	2019	2022	Both 2019 and 2022
Sardinia	111	114	102
Sicily	153	172	148

Note. The Table displays the number of municipalities from Sardinia and Sicily observed in each year under consideration.

Table 2 reports statistics on the variables of interest (arrivals and stays are considered in total, without distinguishing between domestic and foreign tourists). Sardinian and Sicilian data reflect what is known for the tourism markets in general: in 2020, there was a sharp contraction in both arrivals and stays, roughly halved compared to the previous year, due to COVID-19 restrictions. In 2021, with some restrictions still in place, there was a partial recovery; in 2022, all restrictions were lifted, and tourism flows saw significant growth, though not fully recovered compared to 2019.

Table 2.

Statistics.

			2019	2020	2021	2022
Sardinia	Across municipalities	Arrivals	Mean: 30,415 Min: 243 Max: 0.305µ	Mean: 13,794 Min: 123 Max: 0.115µ	Mean: 22,367 Min: 241 Max: 0.210µ	Mean: 29,125 Min: 353 Max: 0.322µ
	Across municipalities	Stays	Mean: 134,470 Min: 899 Max: 1.209µ	Mean: 60,120 Min: 1160 Max: 0.401µ	Mean: 97,369 Min: 1101 Max: 0.795µ	Mean: 126,750 Min 1011 Max: 1.158µ
	Total (regional aggregate)	Arrivals	3,444,058	1,475,520	2,466,091	3,411,892
	Total (regional aggregate)	Stays	15,145,885	6,321,111	10,632,221	14,700,911
Sicily	Across municipalities	Arrivals	Mean: 32,956 Min: 210 Max: 0.726µ	Mean: 14,054 Min: 143 Max: 0.726µ	Mean: 18,985 Min: 271 Max: 0.409µ	Mean: 27,936 Min: 368 Max: 0.722µ
	Across municipalities	Stays	Mean: 97,229 Min: 827 Max: 1.594µ	Mean: 42,182 Min: 462 Max: 0.650µ	Mean: 49,222 Min: 1034 Max: 0.987µ	Mean: 84,657 Min: 1015 Max: 1.649µ
	Total (regional aggregate)	Arrivals	5,120,421	2,206,469	3,113,379	4,888,423
	Total (regional aggregate)	Stays	15,114,931	6,622,498	9,689,251	14,783,156

Note. Max values across municipalities, for arrivals and stays, are expressed in millions (µ stays for million).

Results

Table 3 presents the estimation results of Model (1), for arrivals and stays, considering domestic and international flows separately. The focus is on the estimate of the delta coefficient, which emerges to be statistically insignificant in all cases. The Table also considers a specification in which dummy variables are introduced, associated to the type of destination as classified by ISTAT.⁶ Specifically, we introduced a dummy for sea destinations, and another for the set of mountain, hill, lake, sea, thermal, and art destinations. The treatment turns out to be insignificant, even under such a specification. It is worth reporting that the delta coefficient remains statistically insignificant, even when the regression analysis is limited to the subset of municipalities classified as “Sea destination,” or subgroups such as “mountain,” “hill,” “thermal,” or “art” destinations (results, not included for brevity, are available upon request).

Table 3.

Regression results of model 1.

Dependent variables: Arrivals
	Domestic arrivals	Foreign arrivals	Total arrivals	Domestic arrivals	Foreign arrivals	Total arrivals
Alpha	15009.6** (4.62)	15405.1** (4.31)	30414.7** (4.57)	12056.8** (4.31)	12392.4** (4.12)	24449.1** (4.30)
Beta	2462.2 (0.58)	79.5 (0.02)	2541.6 (0.29)	−3910.5 (−1.06)	−6896.3* (−1.69)	−10806.7 (−1.44)
Gamma	401.0 (0.08)	−1690.4 (−0.33)	−1289.4 (−0.14)	478.0 (0.12)	−1611.8 (−0.37)	−1133.8 (−0.14)
Delta	−1583.18 (−0.27)	−2146.7 (−0.33)	−3729.8 (−0.31)	−212.1 (−0.04)	−617.0 (−0.11)	−829.1 (−0−08)
Sea				63993.6** (10.9)	78471.5** (12.8)	132465.1** (11.91)
Mount, hill, thermal, and art				50682.7** (9.65)	45121** (7.77)	95804.1** (8.96)
R2	0.001	0.002	0.002	0.266	0.263	0.276
Dependent variable: Stays
	Domestic stays	Foreign stays	Total stays	Domestic stays	Foreign stays	Total stays
Alpha	65757.5** (6.73)	68716.5** (5.79)	134474.0** (6.52)	58725.3** (6.56)	60073.4** (5.74)	118799** (6.49)
Beta	−17844.7 (−1.39)	−19400.1 (−1.24)	−37244.7 (−1.37)	−33182** (−2.81)	−40708** (−2.94)	−73891.0** (−3.05)
Gamma	2973.04 (0.22)	−10693.6 (−0.64)	−7720.5 (−0.27)	3156.5 (0.25)	−10468.1 (−0.71)	−7311.6 (−0.29)
Delta	−3099.9 (−0.17)	−1751.6 (−0.08)	−4851.5 (−0.13)	209.6 (0.01)	2995.3 (0.16)	3204.9 (0.10)
Sea				156883.0** (8.35)	261182.0** (11.9)	418065.0** (10.8)
Mount, hill, thermal, and art				118461.0 (7.05)	111419.1 (5.66)	229880.0** (6.66)
R2	0.001	0.001	0.001	0.177	0.238	0.225

Note. Regression result of Model 1 (diff-in-diff model on data in levels, 2019−2022), 553 Observations. t-stat in parenthesis; *(**) denote statistical significance at the 10% (5%) level.

Table 4 reports Model (2) estimation results on joint-repeated data considered in difference, so that individual effects are eliminated. All delta coefficients remain statistically insignificant also in this specification. Furthermore, the statistical insignificance of delta coefficients is robust to the inclusion of dummy variables corresponding to the different types of municipalities. Therefore, the treatment provided in Sicily shows no significant effect, according to all our estimates.

Table 4.

Regression results of model 2.

Dependent variables: Arrivals
	Domestic arrivals	Foreign arrivals	Total arrivals	Domestic arrivals	Foreign arrivals	Total arrivals
Alpha	813,5 (1.58)	−1457.1** (−2.66)	−643.7 (−0.86)	773.3 (1.49)	−1114.9** (−2.12)	−341.6 (−0.46)
Delta	−104.76 (−0.16)	−736.1 (−1.02)	−840.9 (−0.87)	−192.4 (−0.28)	−212.3 (0.76)	−404.7 (−0.42)
Sea				2125.1 (1.31)	−7189.8** (−4.38)	−5064** (−2.19)
Mount, hill, thermal, and art				−27.4 (−0.02)	−5217.3** (−3.80)	−5244.7** (−2.71)
R2	0.001	0.004	0.003	0.007	0.117	0.047
Dependent variables: Stays
	Domestic stays	Foreign stays	Total stays	Domestic stays	Foreign stays	Total stays
Alpha	5136.0** (2.07)	−9982.**4 (−3.67)	−4846.4 (−1.44)	4453.8* (1.81)	−8914.4** (−3.31)	−4460.6 (−1.32)
Delta	395.5 (0.12)	2641.4 (0.74)	3036.9 (0.69)	−792.6 (−0.24)	4247.3 (1.21)	3454.7 (0.78)
Sea				21377.7** (2.77)	−21015.8** (−2.49)	361.9 (0.03)
Mount, hill, thermal, and art				6877.8 (1.07)	−16993.1** (−2.41)	−10115.3 (−1.14)
R2	0.015	0.002	0.019	0.033	0−046	0.007

Note. Regression results of Model 2 (diff-in-diff on joint-repeated differentiated data, 2019–2022), 250 Observations. t-stat in parenthesis; *(**) denote statistical significance at the 10% (5%) level.

Concluding remarks

Our results lead to conclude that Sicily’s policy, centered around granting vouchers, appears to be ineffective in influencing observed tourism flows, no matters how many tourists have benefitted from the promotion. We can reflect on potential reasons for this outcome.

Firstly, an intensive advertising campaign with well-known figures and diverse media channels, both online and traditional, was conducted; hence, poor communication doesn’t seem to be the issue. Instead, the requirement to book through tour operators or agents may have discouraged many potential users. Modern tourists prefer customized experiences, and those who use agents may be less price-sensitive, making price incentives ineffective.

Secondly, the promotion is operative outside the peak season. Typically, people traveling in shoulder seasons are not sea&sun tourists. Once again, these travelers tend to be less price-elastic than the average tourist (see Tisdell, 2013, various contributions in Baum and Lundtorp, 2011; and also Matsuuma and Saito, 2022).

Thirdly, a “one-size-fits-all” policy might not effectively address the different motivations of tourists and their diverse responses to the COVID-19 shock (Alvarez-Diaz et al., 2023; Duro et al., 2023b). Moreover, this policy has neither influenced operators’ strategies nor encouraged product innovation, such as the adoption of digital tools to enable tourists to personalize their tourism experience.

Of course, as a cautionary note, consider that the effects of the incentive might not have been immediately visible and could require a longer timeframe for a comprehensive evaluation. However, though warranting further examination and a longer-term perspective, our analysis demonstrates that the effectiveness of offering price subsidies as a decisive tool to sustain the tourism industry in the post COVID-19 era can be far from convincing.

Footnotes

Acknowledgments

The authors are grateful to the Editor and an anonymous reviewer for helpful comments. The research also contributes to the EU GRInS (“Growing Resilient, Inclusive, and Sustainable”) project. The responsibility for the content rests solely with the authors.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Roberto Cellini

Tiziana Cuccia

Notes

Author biographies

Roberto Cellini is Full Professor of Economics at the University of Catania, Italy, where he teaches Microeconomics, and where he is the Director of the Department of Economics and Business. His primary areas of research include theoretical and econometric aspects of economic growth, with applications to countries and regions, and game theory from a theoretical perspective and as applied to industrial organization, cultural and tourism economics.

Tiziana Cuccia is Full Professor of Economic Policy at the University of Catania, Italy, where she teaches Economic policy and European economic policy. Her scientific interests include the economic evaluation of public goods, topics in cultural economics with focus on the artists’ labor market, and tourism economics with focus on the role of cultural heritage in promoting tourism and stimulating local development.

References

Alba

Popescu

(2023) Romanian holiday vouchers: a chance to travel for low-income employees or an instrument to boost the tourism industry? Sustainability 15: 1330.

Alvarez-Díaz

Chamorro-Rivas

González-Gómez

, et al. (2023) The impact of the COVID-19 outbreak on intra- and inter-regional domestic travel: evidence from Spain. Tourism Economics, in press. DOI: 10.1177/13548166231185422.

Baum

Lundtorp

(eds) (2001) Seasonality in Tourism. Oxford: Pergamon.

Card

Krueger

(1994) Minimum wages and employment: a case study of the fast-food industry in New Jersey and Pennsylvania. The American Economic Review 84: 772–793.

Cuccia

Rizzo

(2011) Tourism seasonality in cultural destinations: empirical evidence from Sicily. Tourism Management 32: 589–595.

Markovıč

Dorčák

Dorčáková

, et al. (2019) Public finance - the impact of recreational vouchers on public finances and support for domestic tourism. In: International Public Finance Conference, Antalya, 24–27 April 2019. DOI: 10.26650/PB/SS10.2019.001.003.

Duro

Turrion-Prats

(2019) Tourism seasonality worldwide. Tourism Management Perspectives 31: 38–53.

Duro

Osório

Perez-Laborda

(2023a) Competition for domestic tourism in the COVID-19 pandemic: a characterization using a contest model. Tourism Economics 29: 378–391.

Duro

Osório

Pérez-Laborda

, et al. (2023b) Are destinations reverting to the pre-pandemic “normal”. Tourism Economics, in press. DOI: 10.1177/13548166231204509.

10.

Matsuura

Saito

(2022) The COVID-19 pandemic and domestic travel subsidies. Annals of Tourism Research 92: 103326.

11.

Tisdell

(ed) (2013) Handbook of Tourism Economics: Analysis, New Applications and Case Studies. Singapore: New Scientific Publisher.

12.

Toubes

Araújo Vila

Fraiz Brea

(2021) Changes in consumption patterns and tourist promotion after the Covid-19 pandemic. Journal of Theoretical and Applied Electronic Commerce Research 16: 1332–1352.