Among the PICTs, tourism is a significant source of foreign currency and local employment. The COVID-19 pandemic shut down most overseas travel, and as of 2022 annual numbers did not returned to pre-pandemic trends. Overseas visitors include sea and air arrivals. It includes excursionists/same-day-visitors i.e. visitors from cruise ships.
Fiji is second, behind only Guam in number of overseas arrivals. Are U.S. active military personnel included in Guam’s numbers? If so then Fiji would be first in yearly non-military visitors: mainly people traveling for holiday, work, or visiting friends and family.
source(here::here("scripts/load-libraries.R"))
theme_set(theme_light() +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank()))
options(scipen = 5)
my_caption <- "Plot: Daniel Moul; Data: South Pacific Community"
my_caption_fbos <- "Plot: Daniel Moul; Data: Fiji Bureau of Statistics via Reserve Bank of Fiji"visitors_all_cat <- read_csv(here("./data/raw/overseas-visitor-arrivals/SPC-DF_OVERSEAS_VISITORS-2.0-all.csv"),
show_col_types = FALSE) |>
clean_names() |>
remove_empty(which = "cols")
visitors_total <- visitors_all_cat |>
filter(overseas_visitors_type_2 == "Total") |>
mutate(place = fct_reorder(pacific_island_countries_and_territories, -obs_value, sum)) |>
mutate(max_annual = max(obs_value),
pct_of_max = obs_value / max_annual,
min_year = min(time_period),
max_year = max(time_period),
pct_of_first_year = obs_value / obs_value[time_period == min_year],
.by = pacific_island_countries_and_territories)fname <- here("./data/raw/rbf/5.4-Visitors-Arrivals-Number-by-Country-of-Residence.xlsx")
visitors_by_country_raw <- readxl::read_xlsx(fname,
sheet = "Table 44",
range = "A2:K43") |>
clean_names() |>
remove_empty(which = "cols") |>
rename(nz = new_zealand,
usa = "united_states_of_america",
uk = "united_kingdom",
europe = "continental_europe") |>
select(-total)
# mutate(m = mean(c_across(x:z)))
visitors_by_country_long <- visitors_by_country_raw |>
rowwise() |>
mutate(total = sum(c_across(australia:others))) |>
pivot_longer(cols = australia:total,
names_to = "country",
values_to = "visitors") |>
mutate(country = fct_reorder(country, -visitors, sum)) |>
mutate(pct_of_yearly = visitors / sum(visitors),
.by = period) |>
mutate(visitors_normalized = visitors / visitors[period == 1983],
yty_growth = visitors / lag(visitors, default = NA) - 1,
.by = country)
# https://en.wikipedia.org/wiki/Fiji_coup
# Also including the 2009 constitutional crisis
# https://en.wikipedia.org/wiki/2009_Fijian_constitutional_crisis
coup_years <- tibble(
period = c(1987, 2000, 2006, 2009)
)Most PICTs experienced a positive trend in annual visitors until the COVID-19 pandemic.
visitors_total |>
mutate(place = fct_reorder(pacific_island_countries_and_territories, -obs_value, sum)) |>
ggplot() +
geom_line(aes(time_period, obs_value,
color = if_else(place == "Fiji",
"purple",
"black"))) +
scale_color_identity() +
facet_wrap(~place, scales = "free_y") +
labs(
title = "Annual overseas visitors",
subtitle = "2005-2022; Y axis varies",
x = NULL,
y = NULL,
caption = my_caption
)
There are many ways one could compare the overseas visitors numbers among PICTs. I offer four below.
p0 <- visitors_total |>
mutate(place = fct_reorder(pacific_island_countries_and_territories, -obs_value, sum)) |>
ggplot() +
geom_line(aes(time_period, obs_value + 1,
color = if_else(place == "Fiji",
"purple",
"black"),
linewidth = if_else(place == "Fiji",
1,
0),
group = place),
alpha = 0.5
) +
geom_smooth(aes(time_period, obs_value + 1),
method = "loess", formula = 'y ~ x',
span = 0.2, se = FALSE) +
scale_color_identity() +
scale_linewidth_continuous(range = c(0.25, 1.0)) +
scale_y_continuous(expand = expansion(mult = c(0.01, 0.05)),
labels = label_number(big.mark = ",")) +
guides(linewidth = "none") +
labs(
subtitle = "A: Count",
x = NULL,
y = NULL
)
p1 <- visitors_total |>
mutate(place = fct_reorder(pacific_island_countries_and_territories, -obs_value, sum)) |>
ggplot() +
geom_line(aes(time_period, obs_value + 1,
color = if_else(place == "Fiji",
"purple",
"black"),
linewidth = if_else(place == "Fiji",
1,
0),
group = place),
alpha = 0.5
) +
geom_smooth(aes(time_period, obs_value + 1),
method = "loess", formula = 'y ~ x',
span = 0.2, se = FALSE) +
scale_color_identity() +
scale_linewidth_continuous(range = c(0.25, 1.0)) +
scale_y_log10(expand = expansion(mult = c(0.01, 0.05)),
labels = label_number(big.mark = ",")) +
guides(linewidth = "none") +
labs(
subtitle = "B: Count (log10 scale)",
x = NULL,
y = NULL
)
p2 <- visitors_total |>
mutate(place = fct_reorder(pacific_island_countries_and_territories, -obs_value, sum)) |>
ggplot() +
geom_line(aes(time_period, pct_of_first_year,
color = if_else(place == "Fiji",
"purple",
"black"),
linewidth = if_else(place == "Fiji",
1,
0),
group = place),
alpha = 0.5
) +
geom_smooth(aes(time_period, pct_of_first_year),
method = "loess", formula = 'y ~ x',
span = 0.2, se = FALSE) +
scale_color_identity() +
scale_linewidth_continuous(range = c(0.25, 1.0)) +
scale_y_continuous(expand = expansion(mult = c(0.01, 0.05)),
labels = label_number(big.mark = ",")) +
guides(linewidth = "none") +
labs(
subtitle = "C: Trends normalized (1.0 = value for each place at first year in data set)",
x = NULL,
y = NULL
)
p3 <- visitors_total |>
mutate(place = fct_reorder(pacific_island_countries_and_territories, -obs_value, sum)) |>
ggplot() +
geom_line(aes(time_period, pct_of_max,
color = if_else(place == "Fiji",
"purple",
"black"),
linewidth = if_else(place == "Fiji",
1,
0),
group = place),
alpha = 0.5
) +
geom_smooth(aes(time_period, pct_of_max),
method = "loess", formula = 'y ~ x',
span = 0.2, se = FALSE) +
scale_color_identity() +
scale_linewidth_continuous(range = c(0.25, 1.0)) +
scale_y_continuous(expand = expansion(mult = c(0.01, 0.05)),
labels = label_number(big.mark = ",")) +
guides(linewidth = "none") +
labs(
subtitle = "D: Trends normalized (1.0 = max for each place)",
x = NULL,
y = NULL
)
p0 + p1 + p2 + p3 +
plot_annotation(
title = "Annual overseas visitor trends: four views",
subtitle = "Fiji = purple. 2000-2022",
caption = my_caption
)
dta_for_plot <- visitors_by_country_long |>
filter(country != "total")
year_min <- min(dta_for_plot$period)
year_max <- max(dta_for_plot$period)
p1 <- dta_for_plot |>
ggplot() +
geom_line(aes(x = period, y = visitors, color = country),
show.legend = FALSE) +
scale_x_continuous(expand = expansion(mult = c(0, 0))) +
scale_y_continuous(expand = expansion(mult = c(0.002, 0.02)),
labels = label_number(scale_cut = cut_short_scale())) +
labs(
title = glue("Count"),
x = NULL,
y = NULL
)
p2 <- dta_for_plot |>
ggplot() +
geom_line(aes(x = period, y = pct_of_yearly, color = country),
show.legend = FALSE) +
scale_x_continuous(expand = expansion(mult = c(0, 0))) +
scale_y_continuous(expand = expansion(mult = c(0.002, 0.02)),
labels = label_percent()) +
guides(color = guide_legend(override.aes = list(linewidth = 3))) +
labs(
title = glue("Percent of yearly"),
x = NULL,
y = NULL,
color = NULL
)
p3 <- dta_for_plot |>
ggplot() +
geom_line(aes(x = period, y = visitors_normalized, color = country),
show.legend = TRUE) +
scale_x_continuous(expand = expansion(mult = c(0, 0))) +
scale_y_continuous(expand = expansion(mult = c(0.002, 0.02)),
labels = label_percent()) +
guides(color = guide_legend(override.aes = list(linewidth = 3))) +
labs(
title = glue("Relative growth (100% = 1983)"),
x = NULL,
y = NULL,
color = NULL
)
p1 / p2 / p3 +
plot_annotation(
title = glue("Yearly visitors to Fiji by country {year_min} to {year_max}"),
# subtitle = "Vertical lines are years with coups or constitutional crises",
caption = my_caption_fbos
) +
plot_layout(guides = "collect")
dta_for_plot <- visitors_by_country_long |>
filter(country != "total",
period < 2020)
year_min <- min(dta_for_plot$period)
year_max <- max(dta_for_plot$period)
dta_for_plot |>
ggplot(aes(x = period, y = yty_growth, color = country)) +
geom_vline(xintercept = coup_years$period, lty = 2, linewidth = 0.25, alpha = 0.4) +
geom_hline(yintercept = 0, linewidth = 0.25, alpha = 0.4) +
geom_point(size = 1.0, alpha = 0.4,
na.rm = TRUE) +
geom_smooth(method = 'loess', formula = 'y ~ x', se = TRUE,
linewidth = 0.75,
fill = "grey", alpha = 0.25,
na.rm = TRUE) +
scale_x_continuous(expand = expansion(mult = c(0.01, 0.01))) +
scale_y_continuous(expand = expansion(mult = c(0.002, 0.02)),
labels = label_percent()) +
guides(color = "none") +
coord_cartesian(ylim = c(NA, 0.55)) +
facet_wrap(~country) +
labs(
title = glue("Year-to-year growth in visitors to Fiji by country {year_min} to {year_max}"),
subtitle = glue("Vertical lines are years with political instability*",
"\nNot surprising that those years saw significant drops in visitors",
" compared to the year prior"),
x = NULL,
y = NULL,
color = NULL,
caption = glue(my_caption_fbos,
"\n*Coup or constitutional crisis",
"\nNot showing one Japanese outlier > 50%")
)
visitors_for_table <- visitors_by_country_long |>
filter(period %in% c(1983, 1993, 2003, 2013, 2023)) |>
select(-pct_of_yearly, -yty_growth) |>
pivot_wider(names_from = period,
values_from = c(visitors, visitors_normalized))
visitors_for_table |>
arrange(-visitors_2023) |>
gt() |>
tab_header(md("**Visitors to Fiji 1983 - 2023 including counts normalized to 1983**")) |>
tab_options(table.font.size = 10) |>
fmt_number(columns = visitors_1983:visitors_2023,
decimals = 0) |>
fmt_percent(columns = visitors_normalized_1983:visitors_normalized_2023,
decimals = 0)Visitors to Fiji 1983 - 2023 including counts normalized to 1983 |
||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| country | visitors_1983 | visitors_1993 | visitors_2003 | visitors_2013 | visitors_2023 | visitors_normalized_1983 | visitors_normalized_1993 | visitors_normalized_2003 | visitors_normalized_2013 | visitors_normalized_2023 |
| total | 191,616 | 287,462 | 430,800 | 657,707 | 929,740 | 100% | 150% | 225% | 343% | 485% |
| australia | 85,027 | 77,609 | 141,873 | 340,151 | 434,533 | 100% | 91% | 167% | 400% | 511% |
| nz | 24,048 | 40,778 | 75,016 | 108,239 | 220,963 | 100% | 170% | 312% | 450% | 919% |
| usa | 25,636 | 42,557 | 58,323 | 55,385 | 99,518 | 100% | 166% | 228% | 216% | 388% |
| others | 4,661 | 8,864 | 21,326 | 48,002 | 54,303 | 100% | 190% | 458% | 1,030% | 1,165% |
| pacific_islands | 10,588 | 16,985 | 28,167 | 39,450 | 54,221 | 100% | 160% | 266% | 373% | 512% |
| europe | 8,330 | 29,786 | 21,847 | 28,905 | 25,921 | 100% | 358% | 262% | 347% | 311% |
| canada | 13,037 | 12,447 | 10,990 | 13,052 | 21,853 | 100% | 95% | 84% | 100% | 168% |
| uk | 5,888 | 20,233 | 49,794 | 17,209 | 10,680 | 100% | 344% | 846% | 292% | 181% |
| japan | 14,401 | 38,203 | 23,464 | 7,314 | 7,748 | 100% | 265% | 163% | 51% | 54% |
visitors_for_table <- visitors_by_country_long |>
filter(period %in% c(1983, 1993, 2003, 2013, 2023)) |>
select(-pct_of_yearly, -visitors_normalized, -yty_growth) |>
arrange(period) |>
mutate(ten_yr_growth = visitors / lag(visitors, default = NA) - 1,
.by = country) |>
pivot_wider(names_from = period,
values_from = c(visitors, ten_yr_growth))
visitors_for_table |>
arrange(-visitors_2023) |>
select(-ten_yr_growth_1983) |>
gt() |>
tab_header(md("**Visitors to Fiji 1983 - 2023 including 10-year growth rates**")) |>
tab_options(table.font.size = 10) |>
fmt_number(columns = visitors_1983:visitors_2023,
decimals = 0) |>
fmt_percent(columns = ten_yr_growth_1993:ten_yr_growth_2023,
decimals = 0)Visitors to Fiji 1983 - 2023 including 10-year growth rates |
|||||||||
|---|---|---|---|---|---|---|---|---|---|
| country | visitors_1983 | visitors_1993 | visitors_2003 | visitors_2013 | visitors_2023 | ten_yr_growth_1993 | ten_yr_growth_2003 | ten_yr_growth_2013 | ten_yr_growth_2023 |
| total | 191,616 | 287,462 | 430,800 | 657,707 | 929,740 | 50% | 50% | 53% | 41% |
| australia | 85,027 | 77,609 | 141,873 | 340,151 | 434,533 | −9% | 83% | 140% | 28% |
| nz | 24,048 | 40,778 | 75,016 | 108,239 | 220,963 | 70% | 84% | 44% | 104% |
| usa | 25,636 | 42,557 | 58,323 | 55,385 | 99,518 | 66% | 37% | −5% | 80% |
| others | 4,661 | 8,864 | 21,326 | 48,002 | 54,303 | 90% | 141% | 125% | 13% |
| pacific_islands | 10,588 | 16,985 | 28,167 | 39,450 | 54,221 | 60% | 66% | 40% | 37% |
| europe | 8,330 | 29,786 | 21,847 | 28,905 | 25,921 | 258% | −27% | 32% | −10% |
| canada | 13,037 | 12,447 | 10,990 | 13,052 | 21,853 | −5% | −12% | 19% | 67% |
| uk | 5,888 | 20,233 | 49,794 | 17,209 | 10,680 | 244% | 146% | −65% | −38% |
| japan | 14,401 | 38,203 | 23,464 | 7,314 | 7,748 | 165% | −39% | −69% | 6% |
https://www.rbf.gov.fj/wp-content/uploads/2024/08/5.5-Hotel-Statistics.xlsx
hotel <- read_xlsx(here("data/raw/rbf/5.5-Hotel-Statistics.xlsx"),
sheet = "Table 5.5",
range = "A2:J38") |>
clean_names() |>
remove_empty(which = c("rows", "cols")) |>
filter(row_number() > 2) |>
rename(
yr = period,
room_nights_available = room_nights,
room_nights_sold = x3,
room_nights_pct_occupied = x4,
guest_nights_overseas = guest_nights,
guest_nights_local = x6,
guest_nights_total = x7,
pct_turnover_to_earnings = hotel_turnover_to_gross_tourism_earnings
) |>
mutate(yr = parse_number(yr),
across(everything(), as.numeric)) |>
# fix units
mutate(room_nights_available = 1000 * room_nights_available,
room_nights_sold = 1000 * room_nights_sold,
guest_nights_overseas = 1000 * guest_nights_overseas,
guest_nights_local = 1000 * guest_nights_local,
guest_nights_total = 1000 * guest_nights_total,
hotel_turnover = 1e6 * hotel_turnover,
gross_tourism_earnings = 1e6 * gross_tourism_earnings
)
hotel_year_min <- min(hotel$yr)
hotel_year_max <- max(hotel$yr)dta_for_plot <- hotel |>
select(yr, hotel_turnover, gross_tourism_earnings, pct_turnover_to_earnings) |>
# mutate(pct_turnover = round(pct_turnover_to_earnings) / 100,
# pct_turnover_y = hotel_turnover) |>
pivot_longer(cols = c(hotel_turnover, gross_tourism_earnings),
names_to = "metric",
values_to = "value")
p1 <- dta_for_plot |>
ggplot() +
geom_area(aes(x = yr, y = value, fill = metric),
alpha = 0.25,
na.rm = TRUE) +
scale_y_continuous(expand = expansion(mult = c(0.005, 0.02)),
labels = label_number(scale_cut = cut_short_scale())) +
scale_fill_viridis_d(end = 0.9, direction = -1) +
guides(color = guide_legend(override.aes = list(linewidth = 3)),
fill = guide_legend(position = "inside")) +
expand_limits(y = 0) +
theme(legend.position.inside = c(0.3, 0.8)) +
labs(
subtitle = glue("Revenue"),
x = NULL,
y = NULL,
color = NULL
)
p2 <- dta_for_plot |>
ggplot() +
geom_hline(yintercept = 1, lty = 2, linewidth = 0.15, alpha = 0.5) +
geom_line(aes(x = yr, y = pct_turnover_to_earnings),
linewidth = 0.5) +
scale_y_continuous(expand = expansion(mult = c(0.005, 0.02)),
# labels = label_percent(scale = 1)
labels = label_number(suffix = "%")
) +
# guides(color = guide_legend(override.aes = list(linewidth = 3))) +
expand_limits(y = 0) +
coord_cartesian(ylim = c(NA, 125)) +
labs(
subtitle = glue("Hotel turnover as a percent of\ngross tourism earnings"),
x = NULL,
y = NULL,
color = NULL
)
p1 + p2 +
plot_annotation(
title = glue("It would be hard to overstate the impact of COVID-19",
"\non the tourism sector in Fiji. The borders were closed",
"\nstarting in 2020 Q1 and ending in 2022 Q1."),
subtitle = glue("{hotel_year_min} to {hotel_year_max}."),
caption = my_caption_fbos
)
dta_for_plot <- hotel |>
mutate(room_occupancy = guest_nights_total / room_nights_sold) |>
select(yr, starts_with("room")) |>
mutate(
room_nights_unsold = room_nights_available - room_nights_sold,
pct_occupied = room_nights_pct_occupied / 100,
) |>
pivot_longer(
cols = c(room_nights_unsold, room_nights_sold),
names_to = "metric",
values_to = "value"
)
rooms_sold_unsold <- tribble(
~x, ~y, ~label,
2010, 2500000, "Rooms unsold",
2010, 800000, "Rooms occupied"
)
p1 <- dta_for_plot |>
ggplot() +
# geom_ribbon(aes(x = yr, ymin = room_nights_sold, ymax = room_nights_available),
# fill = "firebrick", alpha = 0.15,
# na.rm = TRUE) +
geom_area(aes(x = yr, y = value, fill = metric),
alpha = 0.25,
na.rm = TRUE,
show.legend = FALSE) +
geom_line(aes(x = yr, y = room_nights_available),
color = "firebrick", alpha = 0.8, linewidth = 0.15,
na.rm = TRUE) +
# geom_line(aes(x = yr, y = room_nights_sold),
# color = "firebrick", alpha = 0.8, linewidth = 0.15,
# na.rm = TRUE) +
geom_text(data = rooms_sold_unsold,
aes(x = x, y = y, label = label),
size = 3, hjust = 0) +
annotate("text", x = 1998, y = 3.2e6, label = "Total rooms\navailable",
color = "firebrick", alpha = 0.8,
size = 3, hjust = 0
) +
annotate('curve',
x = 2001, y = 3.0e6,
xend = 2005, yend = 2.9e6,
linewidth = 0.5, curvature = 0.25,
arrow = arrow(length = unit(0.25, 'cm'),
# color = "firebrick", alpha = 0.8)
)
) +
scale_y_continuous(expand = expansion(mult = c(0.005, 0.02)),
labels = label_number(scale_cut = cut_short_scale())) +
scale_fill_viridis_d(end = 0.9, direction = -1) +
guides(#color = guide_legend(override.aes = list(linewidth = 3)),
fill = guide_legend(position = "inside")) +
expand_limits(y = 0) +
theme(legend.position.inside = c(0.3, 0.8)) +
labs(
subtitle = glue("Count of room nights"),
x = NULL,
y = NULL,
caption = my_caption_fbos,
color = NULL
)
p2 <- dta_for_plot |>
ggplot() +
geom_line(aes(x = yr, y = pct_occupied),
linewidth = 0.5) +
scale_y_continuous(expand = expansion(mult = c(0.005, 0.02)),
labels = label_percent()) +
# guides(color = guide_legend(override.aes = list(linewidth = 3))) +
expand_limits(y = 0) +
labs(
subtitle = glue("Percent of room nights occupied"),
x = NULL,
y = NULL,
color = NULL
)
p3 <- dta_for_plot |>
ggplot() +
geom_line(aes(x = yr, y = room_occupancy),
linewidth = 0.5) +
scale_y_continuous(expand = expansion(mult = c(0.005, 0.02))) +
# guides(color = guide_legend(override.aes = list(linewidth = 3))) +
expand_limits(y = 0) +
labs(
subtitle = glue("Guests per room"),
x = NULL,
y = NULL,
color = NULL
)
p1 + (p2 / p3) +
plot_annotation(
title = glue("Room nights in Fiji and guests per room"),
subtitle = glue("{hotel_year_min} to {hotel_year_max}."),
caption = my_caption_fbos
)
dta_for_plot <- hotel |>
select(yr, guest_nights_local, guest_nights_overseas) |>
mutate(pct_local = round(guest_nights_local / (guest_nights_local +
guest_nights_overseas), digits = 2),
pct_local_y = guest_nights_local + guest_nights_overseas) |>
pivot_longer(cols = starts_with("guest"),
names_to = "metric",
values_to = "value")
p1 <- dta_for_plot |>
ggplot() +
geom_area(aes(x = yr, y = value, fill = metric),
alpha = 0.25,
na.rm = TRUE) +
scale_y_continuous(expand = expansion(mult = c(0.005, 0.02)),
labels = label_number(scale_cut = cut_short_scale())) +
scale_fill_viridis_d(end = 0.9, direction = -1) +
guides(fill = guide_legend(position = "inside")) +
expand_limits(y = 0) +
theme(legend.position.inside = c(0.3, 0.8)) +
labs(
subtitle = "Count of guest nights",
x = NULL,
y = NULL,
color = NULL
)
p2 <- dta_for_plot |>
ggplot() +
geom_line(aes(x = yr, y = pct_local),
linewidth = 0.5) +
scale_y_continuous(expand = expansion(mult = c(0.005, 0.02)),
labels = label_percent()) +
# guides(fill = guide_legend(position = "inside")) +
expand_limits(y = 0) +
coord_cartesian(ylim = c(NA, 0.3)) +
labs(
subtitle = "Percent of guest nights by locals",
x = NULL,
y = NULL,
color = NULL
)
p1 + p2 +
plot_annotation(
title = glue("An increasing percentage of guest nights are booked by locals",
"\neven as total guests nights has grown in Fiji"),
subtitle = glue("{hotel_year_min} to {hotel_year_max}."),
caption = my_caption_fbos
)