This article by James Alm (2012) examines the studies completedby Michael Allington and Agmar Sandmo, said to be the founders of contemporaryanalysis in the field of tax evasion.
Allington and Sandmo began their researchin this area in 1972, since then exploration within the area has been limited.Due to the limited research, this paper investigates how to describe many ofthe issues revolving around tax evasion, such as defining tax evasion, measuringtax evasion and how to manage the affects using what has been learned from theempirical, theoretical and experimental studies. The article opens by discussing the measurement of taxevasion, and the difficulties that are involved in gathering reliableinformation. The methods of measurement are then broken down into categories, whichfall either under Alm’s classification of “traditional” or “modern.” Within thesecategories are subcategories such as “indirect”, “direct” and “model”approaches which Alm discusses, as well as some alternative forms of measurementused in the more modern approaches. TraditionalApproachesTraditional approaches to measuring tax evasion offermuch more accuracy if they apply a “direct” approach, implying that they arebeing upfront with taxpayers about the study of evasion and analysis of theirtax returns. Examples of direct approaches are the sample taken by the IRS withthe implementation of their “Taxpayer Compliance Measurement Program”, laterreplaced by National Research Program, that audits approximately 50,000 taxreturns per year.
This study used to analyze taxpayer data in a line-by-line manorto estimate the difference between reported income and the estimated actualincome, whereas National Research Program only audit some returns line-by-line.The other direct approaches are less precise than the approach taken by the IRS,as they include surveyed taxpayers, asked to describe their tax evasion conduct.Lastly the analysis of amnesty participants as their tax data was used to derivea measure of evasion based on their income. There are also several “indirect” approaches to measuringtax evasion, these measures are derived by investigating the differences foundin tax return data. These measurements vary from computing differences betweentwo factors, to analyzing the currency market for traces of evasion, to measuringelectricity consumption and comparing it to economic activity. Some of these approachesmeasure the variance between factors, such as reported income and nationalincome, income and expenditure, or the labor force market.
Another “indirect”approach is the currency market, where analyzing currency transactions gives anestimate of the “shadow economy”, this estimate can also be connected to taxevasion as a measurement. Additionally, there has been some analysis of electricityconsumption and how it can predict the level of economic activity which providessome markers of tax evasion that can be found when looking at economic activityestimates and real data. A more limited approach is the “model” approach, as itscharacteristics are, it assumes multiple factors will cause shadow economyactivity, and the multiple effects it will have in the long run. Essentially thisapproach examines several causes and indicators unlike “traditional” or “modern”approaches as well as several indicators. An example of this model is the “DynamicMultiple Indicators – Multiple Causes” model. ModernApproachesOver time the approaches used to estimate evasion haveadapted, many modern approaches today assume individual’s income is split intotwo categories, reported and unreported income. This assumption allows modernmethods to simply consider tax returns and use them to approximate evasion. Othermodern approaches include methods that have conducted field studies, whereasothers have used additional economic data as indicators for evasion such as consumptionand tax deductions.
Similar to the traditional approaches, there is some surveybased data which considers more particular aspects of an individual’s spendingand working habits. The common factor of these measures of evasion is that noneof them include a direct measure of evasion. Modern approaches also include some concepts thatcould be considered a little be out there, for example the concept of measuringthe luminosity of the earth from outer space to approximate economic activityand compared it to income seems like a bit of a stretch. Other interestingmeasures of evasion include cigarette tax evasion. Theoretical explanationsof evasionTheoretical models of tax evasion all use the samebasic model, the economics-of-crime model. This model assumes that rational behavioris to maximize utility of taking a risk, the risk in this scenario is lying orfalsely reporting income on a tax return at the risk of being caught. An adaptationof this model is the “portfolio” model where the risk of evading taxes or aportion of tax at the risk of being caught, which ultimately leads an individualto pay taxes as the fear of being caught outweighs the possible monetary gain.
The”portfolio” model creates the perspective that individuals comply to taxation becausethe risk of audit and or fines is too high. Despite the low audit rate this approachstill uncovers the concept that individual’s compliance is dependent on thepenalties involved. Meaning the overarching theme of this approach is that itassumes that an individual pays taxes solely to avoid the consequences.The avoidance of consequences is an interestingphenomenon in this case, as audit rate is only roughly 1% of all returns, meaningthat the likelihood of being caught is very low or nearly zero. Additionally,the penalties involved in evading taxes are generally mild, as most tax evaderswill simply have to pay the remaining unpaid balance, with penalties infrequentlybeing imposed. That being said, economic theory of rational behavior wouldsuggest that the gamble is worth the risk in this case, as the benefit totaking the risk is available and if you are caught you pay what you would haveanyway.
Therefore, underreporting income or overstating deductions would be a logicalchoice as the proportion of people caught is very low. In spite of this, peopleremain risk averse when it comes to tax. The portfolio model is not without its caveats though,as it tends to suggest individuals should report practically zero income togive themselves the most possible benefit. This model would also predict thatincreasing the tax rate would result in a positive effect on individualsreported income. Despite the economic theory that supports individuals makingthis decision, the compliance rate is in fact much higher than the predictedlevel.
Individuals compliance is consistently higher than predicted bytraditional economic theory, bringing to question the other attributes of compliance.This model is reliant on behavioral economics to have the capacity to predictan individual’s behavior, unfortunately according to the portfolio model,individuals are not acting rationally. Empirical explanationsof evasionThe difficulty with the empirical evidence of taxevasion is the data, as there are no direct measures of evasion and using othereconomic factors as a proxy is the most reliable method available. Despite thedifficulty in finding acceptable data there has been considerable research inthis field, as researchers have found that when a tax rate is too high, individualswill be less likely to comply. These findings oppose that of the portfolio model,but have some theoretical support as economic theory would agree that as taxrates increase so does the appeal to evade taxes, as foregoing taxes wouldprovide benefit at a low risk. Experimentalexplanations of evasionExperimental methods for interpreting tax evasion incorporatetheory in a controlled experiment environment, which can produce more accurate data.
This allows econometric researches to use the data to estimate evasion andother factors, that are not possible to generate in a natural setting, based onthe responses of individuals within the experiment. These factors are why experimentsmay be more beneficial than theory and empirical data. Experiments follow thesame basic design, as individuals in a controlled environment are told to makeas much income as they possibly can. The experiment then precedes to go througha few rounds, as individuals are given a choice on how much they want to report.The experiment does include audits and risk of being caught for underreporting andthe information collected in then analyzed.
The results of these experiments have shown datasimilar to the concepts found in both theoretical and empirical data, as highertax rates lead to less compliance by individuals while an increase audit ratewill lead to greater compliance. These experiments have also discovered that individualstend to exaggerate the prospect of an audit, as the probability of anindividual being audited is much lower than individuals assume.Much like other experimental economics, theseexperiments are not without faults, as many believe the experiments are notrepresentative of the population. Secondly a laboratory test is done in a controlledenvironment where some of the factors implemented by the experimenter may notbe replicable in real world scenarios.
There is also the risk of individuals behaviorbeing affected by the lab environment, as they may act differently since theenvironment is controlled and so are the risks involved. Controlling Evasion Controllingevasion can be done by using information gathered by theoretical, empirical,and experimental models to implement new policy. Alm then argues that there are”paradigms” that can affect evasion with the first being the “enforcement paradigm”.This means enforcing more regular audit and harsh penalties it will minimize evasion.The second paradigm is the “service paradigm” which considers tax authoritysuch as government to facilitate and provide services to taxpayers.
The third andfinal paradigm is the “trust paradigm” which suggests individuals will makeethical, and moral choices to maintain social norms. To conclude, Alm discusses how thisarea requires data from each category. Theoretical, empirical, and experimentalto best measure tax evasion. All the data possible is required to control the plethoraof behavior and motivation that affect it.