Figure 2 below visualised sources of final included studies that meets the review criteria per country across the continents.
A total of 533,324 studies were retrieved across the 14 databases. However, 533,256 were excluded after screening for title and abstracts. After check and scan for a duplicate on the remaining 68 studies, 47 were final eligible ones. Of the final 47 studies, full-text studies were retrieved and assessed against all inclusion and exclusion criteria. In all, 17 studies were admissible after meeting all inclusion criteria, comprising studies that examined impacts or effects of government regulations on behaviour of the population in respect of plastic bags use. Figure 3 provides the PRISMA flow diagram illustrating process and results.
All included studies were published between the years 2007 and 2019. Of the 17 included studies, customer observation (4), customer surveys (8), ministry official interviews (1), data from retailers (3), and some using a mixed methodology (1). On geographical composition, there were studies analysing regulations in Africa (8), Asia (5), and Europe (4). Table 5 provided an outline of the characteristics of the included studies.
Effects of plastic bag regulation by country case
Intervention: A combination of bag ban and levy was introduced in 2007. Bags below 24 microns were banned, and retailers were instructed to set the levy price [21, 22].
Results: In the short-term, consumption dropped by 24% in the weeks after the ban and 50% by 18 months into implementation . Ten years later, the researchers found that 86% of people say that the ban had no impact on their plastic bag use . Further, based on a field survey with an average levy of BWP 0.44, if the levy were raised by BWP 1.00 the consumption would drop by 29.2%. With increased BWP 2.00, the consumption would drop by up to 64% . This indicates that while long term behaviour change may not have resulted if the levies were increased to be more prohibitive, consumption would likely decrease.
Intervention: A ban on plastic carrier bags at retail centres came into force in August 2017 .
Results: After a year and a half of the ban, while it is not clear whether there was an overall reduction in consumption, 49% of survey respondents claimed that they take their own bags while shopping. This presumably has led to a decrease in single-use plastic bag usage. On the other hand, reusable bag ownership has nearly tripled from 4.32 bags to 12 bags per person. Out of this, reusable plastic bag ownership has doubled to 7.7 bags per person. The rise of reusable bag ownership is indicative of the changing shopping culture and is also related to behavioural aspects where shoppers either forget or bring lesser bags than they might need to retail stores. Thus, increased reusable bag ownership does not always translate to actual reuse (Fig. 4).
Intervention: In May 2003, regulation set the minimum thickness of plastic bags issued by retailers at 24 microns and limiting printing to 25% of the surface area of plastic bags (50% for environmentally friendly ink). Additionally, a nominal tax of 46-rand cents (US$ 0.06) was imposed on new plastic bags sold by retailers . However, this only lasted for 3 months, after which due to pressure from plastic bag manufacturers the tax fell, and retailers started charging variable prices .
Results: In the short term (in the first 3 months after the levy), there was a sharp decline in plastic bag demand by approximately 60–90% [25, 26]. However, in the medium term there was a revival in demand for bags coinciding with the removal of the fixed levy. Between August 2003 to June 2004, plastic bag demand varied between 20 and 80% of original consumption levels . Over a 6-year period, from 2003 to 2009, plastic bags per real Rand 1000 of shopping fell by approximately 44%. Out of this, the high-income retailers experienced a 57% decline, and the low-income retailers experienced a 50% reduction . Given the steady increase in plastic bag demand, despite rising nominal and real price per bag, it is expected that consumption will continue to increase, and the plastic littering problem will persist. Overall, therefore, price elasticity of demand has been unexpectedly low in the long term . This suggests that the policy has failed partially. Regulations governing the thickness of plastic bags did not have an impact on consumption and demand .
Intervention: A combination of ban and taxes were instituted in 2010. Bags below 30 microns were banned, and one South African Rand (US$0.10) tax was placed on new bags .
Results: The GRADE rating for this study was low due to study design flaws and lack of standardized presentation of the data collected. However, one finding of the interviews was that consumption of plastic bags reduced for a short time, but shortly returned to pre-tax levels.
Overall assessment for Africa
Botswana and South Africa put in place similar regulatory instruments, such as the proscription of bags below 24 microns and levy on the sale of plastic bags above 24 microns. Zimbabwe also put in place similar regulations with the only difference being bags below 30 microns were banned, and a price was placed on new bag sales above 30 microns. In all three cases, the key finding was that consumption of plastic bags reduced in the short term, immediately following the regulation. However, the impact of the regulation wasn’t sustained in the long term. In Botswana’s case, it is hypothesized that perhaps an increase in the levy will drive down consumption. In South Africa, on the other hand, despite increasing levies, consumption patterns have followed an upward trend. This low elasticity of plastic bags could be due to a combination of factors. The relatively low price of bags in terms of consumer incomes could be a factor. Additionally, low consumer awareness regarding the price of the bags might be contributing to the low-price elasticity . Kenya, on the other hand, adopted a more extreme regulatory system by banning all plastic carrier bags. The GRADE rating for the Omondi and Asari  study is very low and it is unclear whether the ban is applicable to all plastic bags irrespective of thickness. Regardless, single-use plastic consumption appears to have decreased, although the overall impact on plastic use is unknown since reusable plastic bag ownership has nearly doubled.
Intervention: A levy was introduced in June 2008, through a "nationwide regulation requiring all retailers to charge for plastic shopping bags" . Figure 5 shows how nationwide regulation in China influenced consumption behaviour of plastic bags.
Results: Half a year after implementation, "the average number of bags consumed monthly fell from around one million to 0.2 million” . A self-reported consumption reduction of 49% was observed in consumer surveys in two differently wealthy cities. The effect was immediate, though the less wealthy city continues to consume more plastic bags. There are differences among consumer groups. Positive regulation perception correlated with effectiveness. Enforcement is inefficient at informal markets . Substitutes for plastic bags may be missing .
Intervention: A tax was introduced in January 2011 through a consumption tax of US$ 0.06 per bag on Saturdays (“No Plastic Bag Campaign” day).
Results: Studies [31, 32] did not assess the ex-ante consumption, but only the ex-post-consumer behaviour. As such, no variance in consumption can be detected.
Overall assessment for East Asia
Because only the studies reporting on China discuss significant evidence, a broad statement of the effectiveness of regulation in East Asia cannot be made and suggest a further research.
Intervention: A tax of €0.15 was imposed on consumption of plastic bags in 2001 .
Result: The tax was fixed to change consumer behaviour towards the use of plastic bags in Ireland, and not necessarily to generate revenue. This emboldened the Irish government set the tax so high that it would deter consumers from indiscriminate use of plastic bags. The tax of €0.15 was multiple folds higher than the maximum price cited in a survey that was carried out before the levy was implemented, which only 8% of respondents were willing to pay. However, the legislation is said to have recorded huge success in the country, with the positive effect cutting across different stratum of the society including retailers who were not giving plastic bags for free. According to the study, there was a drop of about 94% in the consumption of plastic bags within a few years of coming into effect. Also, plastic bags, which constituted about 5% of litter in Ireland pre-legislation in 2001 dropped to 0.32% of litter in 2002, 0.25% in 2003 and 0.22% in 2004. While there was a small rebound after a few years, probably because consumers got used to the levy, the legislation is still considered a major success.
Intervention: The study was conducted using a proposed levy of NIS 1 (US$0.07) on the purchase of plastic bag . This was an estimation effect study.
Result: Authors stated that 6% of the annual plastic bag consumption was used for outdoor activities, hence could become litter if not properly disposed afterwards. The authors further made the case that even if the proposed levy of $0.07 was increased by 400%, a sizable number of consumers in Israel would still use plastic bags, particularly the people that deploy it for outdoor activities. Hence the litter ratio of 6% might remain stagnant regardless of the levy. These arguments downgrade the author’s view on the proposed tax on plastic bags, and they subsequently suggested that any gain the government aims to generate from imposing levy on the use of plastic bags would be lost in the consumer surplus deprivation that would hit the country. Instead, the study recommended explanatory and educational measures that would encourage reduction in consumption of plastic bags and not impact the consumer surplus adversely.
Intervention: A €0.10 tax on plastic bag consumption which was implemented in February 2015 .
Result: In a bid to reduce the consumption of plastic grocery bags in Portugal and in turn reduce the potential contribution to marine litter, the government introduced a plastic bag tax of €0.10 which was implemented in February 2015. The study analysed the effect of the plastic bag tax on consumer and found out that there was a 74% reduction in the consumption of plastic bags within the first few months of implementation. During the same timeframe, there was a 61% increase in the purchase of reusable plastic bags, but a corresponding 12% increase in the consumption of garbage bags. There was also an insignificant difference between the coastal area and the non-coastal area that were sampled, which shows that the legislation affected the entire country almost equally. Furthermore, supermarkets and hypermarkets offered alternatives to customers including the use of their trolleys to move purchased goods.
Intervention: An introduction of €0.05 levy on the purchase of plastic bags .
Result: The study analysed the consumer reaction to a €0.05 tax on the use of plastic bag in England, and how the success in the legislation laid a foundation for further policies targeting a reduction in environment pollution. The study claimed that prior to the levy, 55% of shoppers in England used plastic bags given by retailers—the number however fell to 21% within 6 months of the implementation. While the study indicated a slight demographic pattern to the consumption of plastic bags, the difference was negligible. The study also projected a reduction in consumption in plastic bags in future comparable to what was applicable in both Wales and Scotland, particularly as there was a comparable short-term reduction in all three countries.
Overall assessment for Europe
From the four studies analysed above, with three focusing on ex-post analysis, it was deduced that taxes on European shoppers as a way of changing behaviour against plastic bags consumption has been considerably successful. There was 82% compliance rate by consumers in the use of plastic bags in Europe. It is also puzzling that while the levy was relatively smaller in England in comparison the other countries reviewed, it had an almost equal effect compared to Portugal. This could mean that the amount of the tax was not the important tool, but the idea of taxing users, and the satisfaction that comes with doing good to the environment. Israel though could be regarded as a deviant case, as the imposition of taxes may not affect the consumption of plastic bags. However, the argument was based on outdoor litter and not necessarily the overall consumption.
Study quality, risk of bias, and quality of evidence
The evidence of quality using GRADE was translated into numeric and depending on the quality category of studies, it is either downgraded or upgraded by points: high (0–1), moderate (2–3), low (4–5) and very low (> 5). Table 7 in Appendix explained the points. The 17 included studies varied in their GRADE scores—two studies received a ‘very low’ score, four received a ‘low’ score, six received a ‘moderate’ score, and five received a ‘high’ score. As outlined in Table 3, seven criteria were used for assessment across all studies, downgrading/upgrading the level of confidence:
The type of design and the inherent likelihood of bias was considered. Randomized customer observation at points-of-sale were considered to be of the lowest likelihood for bias and unstructured interviews with government officials were considered to be of the highest likelihood of bias.
The type of sample (retailer data, customers, ministry officials) also contributed to downgrading based on sample size (the higher, the better).
The funding sources, if any, were taken into consideration for potential influence on study design or analysis.
The degree of indirectness of evidence was evaluated based on how removed the sample type was from the primary objective of consumer behaviour.
The inconsistency of results was evaluated based on consistency of reported findings throughout the study.
The precision of results was considered based on the number of events or participants included in the study and the accompanying specificity of results reported.
Publication bias was assessed based on the presence of selective reporting and justification practices.
Because of the generally overly heterogeneous use of methodology and the lack of control groups and small sample sizes in the selected studies’ design, it was difficult to apply upgrading factors. We therefore restricted the quality assessment to the application of downgrading factors only, giving each one point for a mild limitation and two points for a grave limitation regarding the factors: study design, indirectness, inconsistency, imprecision, and publication bias. This yielded the below ranking (see below Table 6) of quality of evidence in each study.