If you have many cases and are not using a statistical package, you can prepare a csv file in your spreadsheet software, with each line representing a tax return. Taxsim will return a web page with tax liability, etc. for all cases within a few seconds. Accessed via http, Taxsim seems limited to about 2,000 records. This is a browser limitation, not something we do. . Larger CSV files, even millions of records, can be submitted by ftp or ssh or processed on directly or from Stata over the Internet or on your local computer.

For file submissions over http, the first line of your file should show the variable names, using the names in bold on this page. They may be in any order. It is an error to include a variable not named above, but not all variables need to be included. Only year and mstat are required. Fields may be separated by commas or spaces. No missing value codes are allowed - missing data can be represented by zeroes (except for the two required fields).

Here is a sample taxsim input file for two married taxpayers with only 100000 in wages or long term capital gains. Remember missing values must be coded as zero but otherwise the format should be compatible with any generator of .csv files.

The file may include any number of data lines. While the format is flexible, and values may be of variable length, each value must be separated from the next by one or more spaces or a single comma. Tabs won't work. A new case always begins on a new line. We cannot read .prn files, or any file that uses commas or dots to mark thousands. Spreadsheet binary files of any kind can not be read. Before uploading your file please check for the following common errors:

• missing value codes (na, dot, -9, etc)
• out of range state id
• asterisks
• value labels anywhere in the input data
• negative ages, wages, dependents or deductions
• invalid year
• commas embedded in values
• tabs between values
• binary files, (.dta, .xls, .doc, etc).
• spouse income on a single return

The tax calculator stops on any conversion error or value out of range and returns a range error with the case ID.

Marginal Rates (finite differences) with respect to: Other Income Wage Income Taxpayer Earnings Spouse Earnings Long Term Gains Other Deductions Dividends Mortgage Interest Don't bother
Show detailed intermediate calculations: Off On Labeled

The 9 columns of results returned by default are:

• taxsimid = Case ID
• year = Year
• state = State code
• fiitax = Federal income tax liability including capital gains rates, surtaxes, Maximum Tax, NIIT, AMT, Additional Medicare Tax and refundable and non-refundable credits including CTC, ACTC and EIC etc, but not including self-employment or FICA taxes. The adjustment to AGI for SE FICA is made.
• siitax = State income tax liability, also after all credits.
• fica = FICA (OADSI and HI, sum of employee AND employer including Additional Medicare Tax)
• frate = federal marginal rate
• srate = state marginal rate
• ficar = FICA rate
• tfica = taxpayer liability for FICA.

  Marginal rates are with respect to wage income unless another rate is requested. If detailed intermediate results are requested, the following 35 columns of data are added:

• v10 = Federal AGI
• v11 = UI in AGI
• v12 = Social Security in AGI
• v13 = Zero Bracket Amount (zero for itemizers)
• v14 = Personal Exemptions
• v15 = Exemption Phaseout
• v16 = Deduction Phaseout
• v17 = Itemized Deductions in taxable income
• v18 = Federal Taxable Income
• v19 = Tax on Taxable Income (no special capital gains rates)
• v20 = Exemption Surtax
• v21 = General Tax Credit
• v22 = Child Tax Credit (as adjusted includes additional ctc)
• v23 = Additional Child Tax Credit (refundable)
• v24 = Child Care Credit (including additional credit)
• v25 = Earned Income Credit (total federal)
• v26 = Income for the Alternative Minimum Tax
• v27 = AMT Liability after credit for regular tax and other allowed credits.
• v28 = Federal Income Tax Before Credits (includes special treatment of Capital gains, exemption surtax (1988-1996) and 15% rate phaseout (1988-1990) but not AMT)
• v29 = FICA

  The next 11 columns are zeroed if no state is specified:

• v30 = State Household Income (imputation for property tax credit)
• v31 = State Rent Expense (imputation for property tax credit)
• v32 = State AGI
• v33 = State Exemption amount
• v34 = State Standard Deduction
• v35 = State Itemized Deductions
• v36 = State Taxable Income
• v37 = State Property Tax Credit
• v38 = State Child Care Credit
• v39 = State EIC
• v40 = State Total Credits
• v41 = State Bracket Rate

  More federal results:

• v42 = Earned Self-Employment Income for FICA
• v43 = Medicare Tax on Unearned Income
• v44 = Medicare Tax on Earned Income
• v45 = CARES act Recovery Rebates

  More state results:

• staxbc = State tax before credits
• srebate = State income tax rebates (shown only in year paid even if eligibility depends on prior year)
• senergy = State energy/fuel tax credits
• sctc = State child tax credit
• sptcr = Stata property tax credit
• samt = State alternative minimum tax

  Still more federal results:

• qbid = Qualified business income deduction
• niit = Medicare Net Investment Income Tax
• addmed = Medicare additional earnings Tax
• cares = cares rebate
• actc = Additional child tax credit
• cdate = date this version of Taxsim was compiled

In law, the deduction for state tax on the federal return or federal tax on the state return is based on amount paid, not liability. In Taxsim, the deduction is for liability, raising an issue of simultaneity which is resolved by iteration. For example, assuming the taxpayer is an itemizer on both returns, and that they are in a state that allows a deduction for federal tax, then an increase in the wage of a dollar, and that the federal bracket rate is .1 and the state rate is .05, then an additional dollar of wage increases the federal tax by 10 cents in the first round. Then the state tax is calculated and it goes up by .05*.9 = .045. That is a deduction from the federal tax, and so the second round starts with federal income up by only .965, not a full dollar. And so on for 3 rounds.

There are many notches in the tax law. While most are small, they generate implausible marginal tax rates that are probably not well suited for analytical use. TAXSIM uses four strategies to reduce the number of notches affecting marginal rate calculations.

1. Since most notches occur at integer dollar boundaries, we add .1 cent to wages and other property income for every taxpayer.
2. Where step functions are used in the law, we do a linear interpolation along the steps. You can turn most of this off with a tax law modification parameter.
3. The TAXSIM calculation of marginal rates is based on a very small finite difference (1 cent) to minimize the chance of encountering a discontinuity in the tax function, and to make such a discontinuity very obvious if it occurs. For example, a discontinuous dollar jump in tax liability would result in a 10,000 percent tax rate, and be immediately noticable.
4. If the tax rate is larger in absolute value than 100% (federal) or 25% (state) the finite difference is taken again, but subtracted rather than added. Any remaining unusual marginal rates should be called to our attention.

These effective marginal rates are important to us, because one of the most important uses TAXSIM can be put to is calculating correct after-tax prices in models of economic behavior. A problem has been the presence of decreasing marginal tax rates (notches) and otherwise implausible marginal rates in these calculations. We maintain a list of significant notches that we have noticed. Although notches generally involved very small sums of money, they could have disproportionate effects on the results of a linear regression, which is sensitive to the squared difference of any observation from the mean.

All calculations are done in double precision, so the small finite difference should not reduce accuracy. The alternative strategy of taking a large difference to dilute the notch was not a success when tried.

Nevertheless, most problems arise from actual notches in the tax laws. For example, in the federal law the child care credit is an integer percent of child care expenses, depending on AGI. In this case we have used a smooth interpolation between adjacent percentages even though the tax code prescribes a step function. For the intended purpose of TAXSIM, a little smoothing is a good thing. Especially at the state level some credits are just a fixed amount for taxpayers with income less than X. In that case the range for an smoothing isn't obvious and we have not done any.

Sophisticated users may observe that since notches are part of the law, our strategems bias the results towards lower marginal rates. This is true, but we judge the reduction in variance to be worthwhile.

We are very interested in other solutions to this problem, and suggestions are welcome.

If you suspect any case has been calculated incorrectly, please extract that case (or a exemplar, if there are many cases) and run it in individual taxpayer mode above. Examine the output to see if it suggests why the result is a surprise, but if you are still of the opinion that TAXSIM got it wrong, please set the case ID (variable 1) to "-1" and email the output web page to me (as an ascii attachment) with a statement of what you think is wrong. In Stata also set variable idtl to "5". I will get back to you within a day (or two at worst) with an explanation or a fix. It is important that you not send large files. Just send an email with one or two cases, the TAXSIM response and the reason you believe the response to be in error.

We do answer all mail, and we do follow up on queries, but it is important to abide by the guidelines in the previous paragraph. An email message with the statement that "TAXSIM is broken" (unless you are getting no result at all), or even "state x is broken for feature y" will be returned with a request for an example in Internet TAXSIM format as described above. Comments about aggregates not matching public totals are not useful - you need to send an example of a specific return that is off. Once we get that, we will compare our results with hand generated results done from a tax form and any problem will soon be resolved. Comments from users have led to significant improvements in the program, and are almost always resolved within 48 hours. You may wish to try your example in another online calculator - I have a list online

For Stata users Taxsimlocal does all the computation on your own computer, and taxsim.wasm does all the computation in your browser. These are perhaps better solutions for users with confidential data.

• Married folk do not usually file separate returns in the US. There is a tax penalty compared to joint filing in most cases. Single folk do not have income or age for a spouse. This is problematic in survey data, since the marital status information is typically current, but the income information is from the prior year. The solution is to either change the marital status to married (so the return status becomes joint) or convert the record to two individual returns. Otherwise taxsim will fail at the first such record, and subsequent records will have only missing tax data.
• Negative taxes and negative marginal rates are due to refundable tax credits, such as the Earned Income Credit (EIC) and the Additional Chlld Tax Credit.
• The FICA estimate includes both employee and employer tax. Half of FICA is paid by the employer, so the variable TFICA is supplied with the employee share.
• The CARES act recovery rebates use 2020 income in the calculation of the 2020 clawback.