The following article is reprinted from PCWorld.com.
Tax-prep software and services all promise to make the annual IRS ritual painless by asking the right questions and then doing the math. But that doesn’t mean they’ll all arrive at the same bottom line. We ran a tax scenario for a fictional family through five popular tax-prep sites. The results? Each site told us that the family owed a different amount in taxes, and the difference between the lowest and highest bill was almost $2000, or about 10 percent of the family’s total tax bill.
Which site was correct? It’s hard to say. The disparity in the returns (on a total tax bill of more than $20,000) stemmed from the instructions the tax sites provided (or failed to provide), but it also reflected a lack of concrete guidance from the IRS.
The services I used for this experiment were the same ones I evaluated for overall usability: CompleteTax, TaxAct, TaxBrain, TaxCut, and TurboTax. This time, however, I relied on the software’s instructions and help to make decisions about some slightly more complicated (but hardly unusual) tax scenarios.
My test family—let’s call them the Smiths—was a couple with one child and a base income of $100,000, augmented by $1,000 in interest income. They spent $6,000 on childcare expenses, paid $1,000 in student-loan interest and $12,000 in mortgage interest, and made two contributions of $3000 to IRAs.
The Smiths had donated a pile of clothes and household items to charity sometime during the year. They had bought 100 shares of stock on January 1, 2008, for $5 per share, and then sold those shares on December 31, 2008, for $45 per share (lucky them!). Also in 2008 they sold a small vacation house for $179,000 (they had bought the house in 1990 for $80,000).
Purchase and sale expenses on a home
The sale of the vacation home proved a real test for the tax-prep packages. All the sites said the Smiths could deduct costs associated with the sale of the home, but some were more precise about what types of expenses could and couldn’t be included.
For example, most sites made it pretty clear that the Smiths could deduct the $15,000 they spent in realtor’s fees and advertising expenses, but what about the $2,500 they spent on staging and painting to hasten the sale? When it came to specifics, CompleteTax, TaxBrain, and TurboTax either provided so little advice or such a long list of what counted that regarding the extra $2,500 as selling expenses seemed reasonable.
TurboTax was especially broad, stating that even home-inspection reports, geological surveys, and title insurance counted—items that together could easily cost hundreds of dollars each.
Only TaxAct and TaxCut were clear enough about what counted and what didn’t, explicitly stating that the Smiths should not include the extra $2500 as a selling expense.
So what selling expenses are legal? IRS Publication 523, “Selling Your Home,” simply says: “Selling expenses include commissions, advertising fees, legal fees, and loan charges paid by the seller, such as loan placement fees or ‘points.’” Painting is generally not an acceptable selling expense (usually it’s considered a repair, according to Publication 523), and the jury is still out on whether staging is an advertising expense.
Selling expenses on the home weren’t the only problem. Four of the sites (TaxAct, TaxBrain, TaxCut, and TurboTax) told the Smiths to include purchase expenses such as realtor commissions and closing costs, which totaled $12,000, in the original cost of the house. That effectively reduced the profit on the sale, thereby helping to lower the Smiths’ tax bill.
However, CompleteTax’s hyperlinked help for entering the cost of the home simply read, “Your basis is the residence’s cost if you bought it or built it”—not exactly a clear signal to include realtor commissions and closing costs. This is a big reason why the CompleteTax bill was the highest of all the sites. (IRS instructions state that closing costs and realtor commissions can be included as part of the home’s cost.)
The stock sale seemed like a simple transaction—until the questions started coming. Was the sale a “normal sale” or was it a “personal loss”? Was it a wash sale or a Section 1202 exclusion? For all five sites, just the act of classifying the sale involved questions that the Smiths would probably have no way of answering without at least a little strategic advice (or jargon-to-English translations). CompleteTax, for example, asked which one of nine categories the stock sale fell into but offered little explanation of each and no real method for determining which one fit. As if to rub salt in the wound, the site’s vague, four-sentence help box on the subject included the phrase “You would usually know if another choice applies.”
Likewise, TaxCut offered a mind-boggling 17 categories of dispositions to choose from—practically an invitation to err, especially considering the equally vague, four-sentence explanation it produced when I clicked “Learn More.” It didn’t bat an eye when I randomly chose “personal use property.”
Of the five sites, TurboTax did the best job here because its process-of-elimination approach used more-specific questions to narrow down the choices. Perhaps because the Smiths held the stock for only a year, the classification issue didn’t affect the bottom line. But picking a category, potentially the wrong one, without first getting sound advice is not a real-world gamble anyone would want to take.
The five sites also gave the Smiths different advice on what they could deduct for their donation to Goodwill of a big box containing 20 shirts, 40 books, five pairs of shoes, five purses, and an Alf doll.
Three of the sites—CompleteTax, TaxBrain, and TaxCut—simply asked what the value of the donation was and did not provide any method to calculate it. So I entered $400. But two of the sites, TaxAct and TurboTax, include proprietary software to assign “real-world” values to items donated to charity, and their valuations were way higher than I would have guessed—which meant more money for the Smiths.
The tools did not agree on the valuations, however. For instance, 10 women’s long-sleeve dress shirts garnered a $145 deduction with TaxAct’s Deduction Assistant, but only $60 with TurboTax’s ItsDeductible module (for some items, however, TurboTax assigned higher values than TaxAct did).
The end result: The Smiths’ box of donated items was worth $647 if they used TaxAct and $534 if they used TurboTax. The difference is significant considering that being in the 25 percent tax bracket effectively means taking 25 cents off the tax bill for every dollar’s worth of items donated.
The IRS, according to its Publication 526 (“Charitable Contributions”), allows taxpayers to deduct the fair market value of donated clothes and household items in most circumstances. But it offers no guidance on where to find, or how to determine, those fair market values; in fact, page 10 of Publication 526 says, “There are no fixed formulas or methods for finding the value of items of clothing.” For household goods, the IRS goes on to say that “formulas (such as using a percentage of the cost to buy a new replacement item) are not acceptable in determining value.”
This different-answer-every-time problem is nothing new in the tax world. In a similar experiment back in the 1980s, Money Magazine found that human tax pros came up with different results for the same tax scenarios as well. And because the IRS has refused to evaluate the accuracy of tax-prep software in applying tax law (saying that would be “a monumental challenge”), the only way you’ll probably ever know if your tax software steered you wrong is if the IRS decides to audit your return.
Tax-preparation software has indeed helped millions of Americans do their taxes more quickly and more cheaply, but these examples illustrate how the bottom line can be directly affected by the amount and quality of hand-holding provided. If you need an entire subset of specialized knowledge and calculations just to fill in one little white box, you might consider calling a CPA—and even then, be prepared to be second-guessed if you’re audited.