Minimum Legal Drinking Age (MLDA) and the Reduction of the Proportion of Drinking

Zheyi Zheng
Econ 104
May 30th, 2019
Minimum Legal Drinking Age (MLDA) and the Reduction of the Proportion of Drinking
Population, Crime Rates, and Arrests
Abstract
The Minimum Legal Drinking Age and Crime (MLDA) is a policy in the US that sets the minimum age where people can legally purchase and consume alcohol as 21 years . This policy limits people below 21 years getting access to alcohol. The analysis focuses on whether the MLDA, affects arrest and crime rates using data from the National Health Interview Survey (NHS). The data includes other demographic variables and help s to compare alcohol use patterns under and above 21 years. Graphs, regression including and the Delta method were used to organize and analyze the data. Based on the regression results, the MLDA reduce the arrest rates of people under 21 years old and the effect on crime is statistically significance with the crimes. There were 82 per 1000 less crimes committed by individuals over 21. The IV assumptions were met and the policy reduced the crime rates and arrests.
1. Introduction
The Minimum Legal Drinking Age and Crime (MLDA) is a policy and type of social regulation in the US that specifies that people are to buy, and consume alcohol legally when they are at least 21 years of age (Toomey, Nelson, and Lenk 1958). The policy is linked to the social belief that alcohol is more harmful to young people and places them at risk. Alcohol consumption is a public health problem that requires immediate preventive actions and health promotion. According to Carpenter and Dobkin (521), harm and damages linked alcohol consumption cost more than $50 billion each year. When there is harmful or undesirable behavior in society the state implements different types of actions: such as economic regulation through tax to discourage behavior and / or correct negative externalities, am information campaign to create awareness and change behavior, and another action is coercive measures to prohibits certain behavior and implement laws.
To determine whether there is a link between MLDA, arrest and crime data sets on alcohol consumption from the National Health Interview Survey (NHS) and arrest rates for California are used. The NIHS database includes demographic data and outcomes for different groups. Using this information it is possible to analyze how the minimum legal drinking age has affected overall drinking, arrests, and crime rates. Using instrumental variation (IV) approach that compares the rates of arrests for young adults who have just turned 21 years old with those ones who are about to turn 21.
The MLDA reduces the rates of alcohol consummation and crime rates and mortality rates. Using the IV estimation, drinking alcohol was associated with higher crime and death rates and since the IV and the MLDA are valid instruments, the regression results are reliable. Crime reduced by 8.2% for the individuals above 21 years, and indicates that changing the legal drinking age would also affect the crime rates. Prohibiting the sale or serving of alcohol limits access to the alcoholic beverages and reduces the risk of engaging in risky behavior.
2. Data
The NHS data are a subsection of the CDC, and helps to understand health trends with the datasets to investigate how MLDA affects the crime rates, and how the policy reduces the mortality rate. The individuals in the NHS are randomly selected and report about their health and lifestyle, and since there is randomization they represent many of the people and households. Alcohol consumption varies by age. The data is classified according to the demographic variables, for which the data has been collected the data for several years making it easier to determine the trends and patterns.
Regression estimates based on the demographic variables focused on differences for the surveyed population just under 21 and after 21. There were changes in patterns 21 years, between the two groups with individuals more likely to drink when they were 21 years and above. There were dummy variables to further determine the patterns and these were: uninsured, HS diploma, Hispanic, white, black employed, married, working, going to school and male .
The effect of age drinking rates and crime rates was also considered where different causes were identified with age centered at 21, and the age profile of arrests for all causes, including violating the liquor laws was and the arrest rates. This was part of the second data set on the causes of arrest included the number of days to age 21, DUI, liquor laws, robbery, aggravated assault, Ot-assault, all crimes combined, the drunk-risk as well as combined-oth.
The datasets are useful to estimates effect of MLDA on crime rates and inclination to drink.
3. Methods
To analyze the results, different methods were used to organize and interpret the data once all the data regarding the variables has been obtained. We started with graphs that were made visible to reflect the desired information and this was followed by regression and the Delta method. The regression equations were used to estimate drinking behaviors and crime rates near the MLDA. The regression specification chosen for this analysis was the cubed polynomial regression design since it is the only regression, which can express the trend, and helps fit the trend for the scatter plots. Furthermore, regression was used to determine the drinking rate patterns and crime while using the IV approach, in which we estimate the proportion of people who drink to mortality rates. Using both graphs and regression equations was relevant to evaluate the effects of the MLDA.
This delta methods was used to estimate the standard errors for every for estimate. Several assumptions are considered in this method as follows. The first assumption in IV is that the first stage in the analysis is non-zero and that the assumed restriction age, which is 21 years, minimizes crime rates. With this assumption, the first stage being non-zero was so consistent with the IV estimation. Variations induced by the adoption of minimum legal drinking age are used in estimating the impact of MLDA policy on the reduction of the proportion of drinking population, crime rates, and arrests. The use of IV has been successful in studying the causal effect of alcohol consumption. According to Lousdal (1), IV method is widely used within the field of economics in inferring causality in the presence of unmeasured confounding. Among the commonly used instruments include, alcohol taxes, alcohol policies, and minimum legal drinking ages. In addition, the second assumption was the use of those people under the age of 21 years.
Since none of the people could be assigned as 21 years, it therefore followed that there was no bias between the different groups and the data of the people aged 21 years were randomly assigned.
Figure 1: Alcohol Drinking Rate using different bin sizes
For Figure 1 above, the bin widths that we used for my age profile of whether or not people drink alcohol is turning to 21 years old were: day, 4-day, 20-day, 30-day, 40-day and 100-day. Turning to 21 is the midpoint of the bin widths and the bin will separate from 21 but not across this point. I used 40 days as the bin width, for the reason that we can divide these people who already 21 and who are turning to 21 into two different areas. We can also recognize that all of the bin will across to 21 and make these graphs hard to recognize in other graphs. However, when the bin size equals to 100, there are too few bins in the graph and this graph cannot show any thing. Hence, I chose the bin width with 40 days where the sample size is relatively big enough to observe the trend.
Figure 2: Figure 1: Alcohol Drinking Rate with different band sizes
Figure 2 above, shows the bandwidth sizes are 4-year, 2-year, 1-year and half year representing the age profile of whether or not people drink alcohol. I choose band width of 2 years since the points do not cross the threshold and the graph is clear compared to the 4-year option where the points are dense and almost indistinguishable, and for 1 year and half year the points are too dispersed. As the size of bandwidth becomes smaller, the graph dots are more visible and band size of 2-year has a clear line and dots.
The overall a crime rate of the population per 100,000 individuals and drink rate are the outcomes. The MLDA combined with the data on crime rates and arrests estimate the causal effect of drinking on the crime rate. The IV estimator estimate the effect of alcohol consumption in different stages and the Delta method estimated the standard errors standard errors for deaths.
4. Results
Figure 3: Arrest profile all cases
Figure 3 above shows the age profile of drinking fitted with the regression line.
Table1: Balanced table that shows that none of the other covariates are changing sharply at age 21.
Variables

-1

-2

-3

-4

-5

-6


drinks-alcohol

drinks-alcohol

drinks-alcohol<...