Maybe a Walmart near your home isn’t so bad after all

From Devin G. Pope and Jaren C. Pope:

Walmart often faces strong local opposition when trying to build a new store. Opponents often claim that Walmart lowers nearby housing prices. In this study we use over one million housing transactions located near 159 Walmarts that opened between 2000 and 2006 to test if the opening of a Walmart does indeed lower housing prices. Using a difference-in-differences specification, our estimates suggest that a new Walmart store actually increases housing prices by between 2 and 3 percent for houses located within 0.5 miles of the store and by 1 to 2 percent for houses located between 0.5 and 1 mile.

Does anyone know of an ungated copy of the paper?


Sorry, here is an ungated version:

I'm not sure if this is a well-formed question, but are these percent increases median or average?


It's terrible! Noone can afford to live near a Wal-Mart and have access to food and clothing and toys and dish detergent and...!

Not only does Wal-Mart provide affordable necessities to the poor by the trainload, they do it by paying good wages to over one million non-union workers.

It is for the latter that they must be destroyed.

Not having read the paper, but my hypothesis would be that it could well have a net-positive effect on home prices, but that it would not be a consistent effect. At the lower end of the market, a nearby Walmart would raise house values because of the access to a highly-useful amenity for lower and middle income people; at the higher end, it could still lower values.


I'm a big fan of Walmart since I have lived in rural areas and my work causes me to spend a lot of time in the back of beyond where Walmart has made life so much better. However in a wealthy neighborhood I strongly suspect that a Walmart would have a very negative effect, if for no other reason than it would draw poor and lower middle class people into the area. This would certainly lower housing values. So Walmart in Elko or Fallon, Nevada is a huge plus. Walmart in Spokane Valley or Hayward, CA can be a neighborhood amenity. In Edina or Shaker Heights, it could be a negative, and in Bel Air or Bloomfield Hills it would be a disaster for local property owners.

I had the same thought. I looked through the paper to see if they addressed this in detail, and it appears they did not, unless I missed something. They offer a description of the average home in their study ($267,000, 1,767 square feet, 2 baths, 30 years old, 0.25 acres and 3 bedrooms), but they don't break out the effect of a nearby Walmart on different tiers of house prices.

However, a home's value and its distance from Walmart appear to be related. The houses closest to a Walmart get the highest Walmart-related increase in value, but still have the lowest overall value. Houses farthest from Walmart get the smallest Walmart-related increase in value, but also have the highest sale prices. One is tempted to draw the conclusion that Walmart doesn't increase the value of property, but rather increases the value of cheap property - which, as other commenters have pointed out, is likely a function of Walmart opening stores primarily in lower income areas.

I wonder how the authors of the paper, or Tyler for that matter, would feel if a Walmart opened up within a half mile of their homes. I'm guessing they wouldn't be terribly happy about it.

Does Wal Mart open stores primarily in low-income areas? I'm pretty sure that was true initially, as they focused on relatively rural areas, but in the last 10-15 years, at least where I've been living, they seem to have been moving primarily into wealthier areas. When I lived in the Dallas area, in a relatively wealthy suburb, there were 7 Wal Marts within 10 miles of my house, only one of which was in what I'd categorize as a low income area. Indeed, before I moved, the newest Wal Mart in the area had been built at the intersection of Park Blvd and the Dallas North Tollway, which is an extremely wealthy area.

@kiwi dave: So in other words, it has the indirect effect of transferring wealth from the 1% to the 99%. Should make Occupy Wall Street very happy.

I think it's much more likely to be a transfer of wealth within the 99%.

They compared the 2.5 years before the Walmart opened to the 2.5 years after the opening, but there was a housing bubble during this period. Housing prices continued to rise at a faster rate. Wouldn't a better control group have been similar communities where no Walmarts were built nearby?

A lack of counterfactual data (i.e. Walmart wanted to build but didn't/couldn't) was an initial concern of mine as well but a housing bubble doesn't explain the effect of geographical proximity on prices.

Of course it does. See below.

It doesn't appear that you understand the statistical model from what you wrote.

Very interesting paper. No obvious holes from a quick skim like some papers that get posted here. I really like the data work and methodology.

Without the paper available, it's hard to judge, but this is highly doubtful.

First, the model likely suffers from specification bias because there is some exogenous factor that affects both the decision to open a Walmart and house prices. Walmart conducts a market analysis before such a decision, and they wouldn't put a store in an area where the economy was in a downward trend.

If Walmart has any effect, it serves as an anchor for other stores in the strip, so there's a spillover effect. The entire strip serves as the amenity. Commercial strips enter areas AFTER home development, and home prices (especially during the subject period) were at their lowest when they first hit the market. We had a commercial boom AFTER the housing boom.

The model also uses a period of time where house prices were almost universally rising at a fast pace. Substitute McDonalds or a Shell gas station and you'd likely see the same effect. It would have been more meaningful if they hadn't picked a trending market but rather looked through an entire business cycle. This data is certainly available.

Why would distance matter so much when everyone DRIVES to Walmart? Some areas are conducive to Walmart development and others not. A house five or ten miles from a Walmart might enjoy the amenity just as much as a house less than one mile away. The conveniences of Walmart are its large lots and one-stop shopping. Walmart is going to place a store where property values or rents are cheap, and who wants a house within a mile of that? Again, the generally rising economy and soaring house prices takes over and ANY business within a reasonable radius could take credit for the increase. During the housing boom, low priced houses appreciated faster than higher priced homes. So this is not a Walmart effect.

I suspect there is a Starbucks effect, too, under the same methodology.

Usually step 1 is to read the paper and then write 4 paragraphs on potential problems. All of your concerns are addressed in the paper either explicitly or implicitly.

If you understood that the underlying trend in the economy that you are so concerned about is differenced out, half of what you wrote goes away.

Usually, Step 1 is to have the paper available to read. After I posted, I saw the ungated link to the paper and read it.

Now that I've read it, I see that they addressed many of the concerns I had. The authors are smart guys.

I see how they differenced away time-invariant omitted variables. I'm less convinced that they've sufficiently tested or controlled for the general trend in house prices during this era nor the endogenous decisions by Walmart on where to place locations. They clearly recognized that these problems exist.

I'll take another look at this later when I have more time.

"However, we must rely on the identifying assumption that housing price trends
for areas near the Walmart and those areas slightly farther away from the Walmart would
have been the same had the Walmart store (and any other stores from the agglomeration
effect) not been built."


That's the problem with a "quick skim."

The main problem is twofold:

1) The paper focuses on 2000 -2006. That seems a relatively small number of years for an effect espeically with absurd housing increases. Stretch the study out to 1980 - 2010. Why not do 30 years or so?
2) The causation is backwards. Wal Mart could have been using housing housing prices to find where to put new stores. You could replace the study with Hollywood Video, Circuit City, Staples, Best Buy, or Costco and probably get the same result. That proves increased commercial real estate increases housing prices.


Why do houses closer to Wal-Mart appreciate faster than houses farther away after the opening? In your model are firms really able to first perfectly identify and then freely locate in the fastest-growing zip code of their target area?

Seems to conflict with the "Wal-Mart goes to where rents and land costs are cheapest" hypothesis.

Why would half a mile or three miles matter?

The paper says that 42% of Americans shop at Walmart regularly. Do 42% of Americans live within 4 miles of a Walmart? What is the average distance a Walmart shopper travels?

The distance matters because the surrounding land is relatively cheap and during the boom cheap houses appreciated faster than more expensive houses because of the availability of financing. But people many miles away are Walmart shoppers.

The main point I was trying to make is:

1) Considering the short time frame home prices probably highly correlated to population increases which I am sure Wal Mart follows very closely. That is why I would want a longer time frame. 1980 - 2010.
2) In general it makes sense but replace Wal Mart with retail center. The authors I sure wanted to prove it was Wal Mart causing this, but I would believe any popular retail outlet this would be true as well. If they wanted to prove Wal Mart then they needed a control group, ie Target, Blockbuster, Best Buy, etc.


Going back to 1980 would be helpful if data were available, but the authors had problems with some data. The data in this paper doesn't include a complete business cycle. Even going back to 1990 there was considerable appreciation, and even faster after 1996.

Identifying acyclical markets and using only that data would be a good control.

As you say, effect may precede cause if Walmart has superior forecasting ability. They can locate stores where development is just beginning. Their doors may open before the first house is sold.

Wal-Mart is comparable to a Target, but not a Best Buy or Staples. Although these are often found in the same vicinity, shoppers are much more likely to show up and spend frequently in a store like Wal-Mart, Cosco or Target than they are in Best Buy or Staples. (My personal ratio for visits is about 5:1, and the spend is 10-12:1).

Having said that, the value of a Wal-Mart versus a normal grocery store is probably not more than $1,000 per year. Capitalize that, and you're probably talking not more than $10,000. That's about 2% of a $500k house. On that basis, the authors' estimates looks reasonable.

Intuitively, the result seems reasonable. People prefer to live near easy shopping (although perhaps, not TOO near), and lots of people shop at WalMart.

I don't understand how a Walmart half a mile away would LOWER prices, unless there are traffic issues or other externalities. I'd imagine there are some people who hate Walmart on principle, but it doesn't seem like they actually suffer any non-psychic loss from having one nearby. Whereas there are strong and obvious gains for the people who *do* shop there.

And I think "traffic issues" are another thing people like to cite, but that don't really have any real bearing on whether one buys a house (barring actually being on a main thoroughfare).

I don't care enough to look into it but I want to point out that it's possible that what attracts Walmart to build a store in a particular location correlates with rapidly rising housing prices, such that the effect of the Walmart is negligible or negative.

I also imagine the effect is a function of the housing prices. It's hard for me to believe that a neighborhood with median house prices > $500k is getting a significant boost, but I would not be so surprised that a neighborhood with median house prices < $100k would.


This was exactly my first reaction.

I didn't have any prior that a Walmart would lower housing values, but there can't be a lot of comparable sales of houses within .5 miles of a Wal-Mart built between 2000 and 2006 I would wonder whether that result was sensitive to a small number of atypical locations.

I agree. I had no priors that a Walmart might attract crime or otherwise cause negative externalities. It seems like they were hedging against an unexpected coefficient sign or arguing the result is stronger because the impact was ambiguous.

Zoning would limit the number or houses in that radius. Since their coefficients were significant, the small sample size didn't seem to affect the test results.

I'm not suggesting their coefficients were not significant; I'm wondering whether, given the small sample, the significance might come from a few specific observations with unusual characteristics. I have a hard time understanding why you would see a larger increase for a distance within half a mile vs within a mile, given that virtually all trips to Walmart are made by car.

Actually, how many houses are there typically in a 0.5 mile (2600 ft) radius of a Walmart? An average Walmart store itself is 110,000 square feet. I think zoning codes mandate somewhere around a 1:3 ratio for floor space to parking lot space, so add in a 330,000 square feet parking lot.

This combined area translates to a equivalent radius of about 700 feet. Add in zoning buffer zones, and factor in the fact that most Walmarts are located in fairly industrial / commercial zones.

So I'm curious how many houses at all are located within the 0.5 mile radius and of them only a small fraction end up as sales? Is this a pretty thin data-set? What's the typical number of transactions found in this zone by the authors?

Let me summarize the paper in two sentences:

Before Walmart, nearby houses sell for 5% less than expected. After Walmart, 2.5% less.

The moral you draw from this depends on your biases. Mine tell me that Walmart locates in low-cost neighborhoods, and that less-wealthy homeowners are the main beneficiaries.

This is the first time I've ever heard it suggested that Walmart lowers housing prices or increases crime and congestion, in contrast with the much more common critiques that it drives down wages and benefits and destroys small retailers. So beyond any specific problems with the analysis this feels a bit strawmanish.

Exactly. And Rahul's point is connected - how much do the positive effects (shown in the paper) outweigh these potential costs ? That is what a policy relevant cost benefit analysis needs.

I don't understand these techniques much; but my naive question is how significant are these results? 1-2% increases seem small. If you were to compute the effect of random geographical artifacts on house prices, say a new traffic light etc. would you expect the data-set to yield a number very close to zero?

The other metric that I didn't see during my skim of the paper was some sort of a error bar. 1-2% average increases but what's the variation like in the case of individual Walmarts?

You're asking a good question between statistical vs. economic significance. The tests show statistical significance. A traffic light is a public good, so not quite comparable to the private service but you are correct that any single hedonic property is going to add a small amount to house price variations. Given thousands of hedonic factors, 2% is a large single contributor. It would be more plausible if Walmart was an anchoring business with positive spillovers to commerce.

If the variable is significant at five percent, then a 95% confidence interval would not contain zero, assuming they did a two tailed test. Their statement about ambiguous priors suggests they did. Therefore the standard error times (about) 2 is less than 1%.

Oh, I didn't mean a traffic light in its sense as a public good. I used it as a random geographical artifact. You could use, say, the location of new water-fountains, or pet-kennels or day-care centers instead.

My point is, if I analysed the same house-transaction date with respect to 0.5-mile radii of trivial geographical markers how consistently close to a 0.0% house-price increase am I going to get? In that sense, how significant is a 1% increase?

What where these Walmart's before these stores opened? Undeveloped commercial land? Farmland? Some other form of retail that went under?

I'm glad I read the paper because a lot of the issues were addressed in their methodology. This doesn't mean they were resolved but the results don't seem out of line.

Pope and Pope are brothers?

Can someone please help me understand the following statement:

"For example, if Walmarts tend to be built in areas where there is higher
crime, then a cross-sectional estimate of the implicit price for living near a Walmart that excludes the relevant measures of crime will be biased downwards (more negative)."

Comments for this post are closed