Do Government Statistics Yield Better Business Outcomes?

>> Steven Davis: Article 1, Section 2 of the US Constitution requires an enumeration of persons residing in the United States every 10 years. The federal government meets this requirement through its decennial Census of population and Housing. It uses the census data to apportion members of the US house of Representatives among the various states.

 

And importantly, to allocate funds for government programs to states and communities. As it turns out, these data are also quite valuable in the private sector for commercial purposes. That's the topic of today's show. Welcome to Economics Applied. My name is Steven Davis, host of the show. Today's guest is Feng Chi.

 

She's a PhD candidate at Cornell University and will join the University of Illinois as an assistant professor of economics later this year. Welcome, Feng.

>> Feng Chi: Thank you for having me, Steve. It's great to be here.

>> Steven Davis: Great to have you. First, I should say congrats on your appointment at the University of Illinois in a tough job market for PhD students in economics.

 

And hats off to the folks at U of I for their good judgment in hiring you. So, I think part of the reason they hired you is that they saw your job market paper, so to speak, growth of your PhD thesis and liked it, as did I. And we're gonna talk about that, the research in your PhD thesis today.

 

So let me just give a little bit more background to set things up. So the decennial census that I mentioned at the outset, that yields detailed information about the local population characteristics of cities, towns, neighborhoods and even smaller geographic areas. And it has information like what's the age distribution of the people living in that area, the family sizes, the education levels, household incomes, ethnic backgrounds, and the like.

 

So there's a lot of detailed demographic and socioeconomic information at a highly granular local level in the decennial census. So explain to the audience how that type of information can be useful for business decision making.

>> Feng Chi: Yeah, absolutely. So demographic information is important for business decision making because it really helps answer two key questions.

 

What are the customers and where to find them? For many businesses, they need to understand demographic trends in order to plan production. For example, if you are a baby product maker, you make formula diapers. So you will really want to find out how many babies are born each year.

 

And if you're a car manufacturer, trends in things like family size will help you decide what type of vehicles that you prioritize producing. Should you make more compact cars or more minivans?

>> Steven Davis: Right, so don't put the kids nursery store in a retirement community.

>> Feng Chi: Yeah.

>> Steven Davis: That's probably not gonna work out very well.

 

 

>> Feng Chi: Yeah.

>> Steven Davis: So it's a somewhat tried example, but it in fact is the kind of thing that retailers in particular need to be attentive to when they decide where to site their businesses. And also what kind of products to stock on their shelves.

>> Feng Chi: And as you mentioned, it's not just about the general demographic trend.

 

Those businesses also want to know exactly where those customers are, where to find them. You probably could get a good sense of what the data says about your neighborhood just by looking at the kind of junk meals you typically receive. Sometimes I get those two for one breakfast coupons from fast food chain.

 

This is great for a graduate student, but sometimes I also get ads in Spanish even though I don't speak the language. So these type of promotions are really targeting demographic profiles at the neighborhood level, not for individuals. And if you get a lot of flyers about roofing or landscaping, it probably means that census data says there are more homeowners than renters in your neighborhood.

 

 

>> Steven Davis: These are great examples. One more, just suppose you have a restaurant in a local area and you want it to be profitable, at least if you want to be economically viable. In addition to thinking exactly where you want to locate it, which you might decide just by driving around, you have to think about, well, what kind of food should I serve?

 

And that's gonna depend on the ethnic makeup of the neighborhood and the age distribution. And do you want an ambiance that appeals to families with kids or young singles, married couples, retirees? How did you price things? All of these very practical business decisions are informed by the demographic characteristics of the surrounding neighborhood or neighborhoods.

 

 

>> Feng Chi: Yeah, so that's where I was going next. Because these local demographic information is especially important for like a retail and restaurant businesses because those businesses really depend on demand from local areas. It's important for them to pick a location that has the right demographic profile. For example, if you are a high end boutique store, you probably will thrive in the high income neighborhood but fail in like a lower income neighborhood.

 

But for dollar store the situation is probably the opposite. So these are not good or bad neighborhoods per se. You want to find the right fit. So that's the kind of business decision I try to study in this paper.

>> Steven Davis: Okay, yeah, finding the right fit for your business, your product selection, your services for the local that meet the local demographic characteristics.

 

Okay, so that's really important for business success. That seems at one level bright and commonsensical. But why can't businesses just get their data in other ways? Why do they have to rely on government sources? Or what's better about government sources? Or how do you even know the government sources are important?

 

 

>> Feng Chi: Well, so a key feature of the census data is it really tries to account for everyone in the nation, so it's conducted at a very large scale. That kind of scale is hard for private sources to match. If you want to look at the local characteristics for a very small area, which is important for some of the retail businesses.

 

You really want to have enough sample size to give you an accurate estimate on how things are in this area.

>> Steven Davis: Okay, so the census is much larger and more granular than other data sources. But I suspect if we went out and surveyed a bunch of, say, small business owners in retail, restaurants, lodging, personal services, that quite a few of them would say, well, I've never used census data.

 

I get my information from my buddy who has a small consulting firm, or I just drive around. Or how do you respond to that? How do you know what's so special about the government statistics, and how do you know?

>> Feng Chi: Well, even for those small businesses, they might actually be using census data without realizing where the data comes from.

 

So even if you're a small business, when you go out driving around and try to look up for a great spot for your next restaurant. You will typically hire a commercial real estate agent that shows you if this area has the kind of demographics that will be great for your business, so that they will charge you a higher premium, say.

 

And also when you apply for loans from banks to fund your business, they typically require you to draft a detailed marketing plan to answer what type of customers you're targeting and what's the potential size of this market. And that's when even small business owners need to look up this information.

 

And you can have access to this information at local public libraries, which typically have experienced staff and small business centers that will help entrepreneurs fill up the information and make decisions.

>> Steven Davis: Okay, so I think there's a couple key points there that I wanna highlight. One is sometimes individual businesses actually do use this information directly, as when they go to the local library and they tap the census data about their local community.

 

But other times, you made an important point here, which I know to be true from my own experience. There are many information service companies out there and consultants who sell information in easy to use forms to business owners or business entrepreneurs, prospective business owners. That actually comes ultimately from the Census Bureau or from other government statistical sources.

 

I can see this because I've used these data. I know many of them quite intimately, so I can see when they're drawing on government census sources. But if you just go into some business consultant and you ask them for advice or information about the local neighborhood, you may get it from the private sector party, but ultimately it's coming in part from these government statistics.

 

Is that right?

>> Feng Chi: Yeah, so like I mentioned before, so the scale and detail of census data is really hard for private sources to match. Say, a big player as an information intermediary in this space is Sri. So they're the creator of the ArcGIS software. So they provide maps and they embed census demographic data into their software.

 

So if you're a business, you only need to pull up the map and filter and choose variables you want to focus on. Say if you want a median income within a certain range, like a population of certain size, you just do a bunch of things in the software, but you don't know where the information comes from.

 

And that information comes from census.

>> Steven Davis: Yeah, so a couple things just to make sure everybody's following. GIS stands for Geographic Information System. And what GIS software does for us is present information rich maps. So I don't know exactly what Google Maps does. But if you look at the search facility on your phone, it'll often tell you the businesses in the local area and so on.

 

So that's an example of a information rich map. And what you're telling us now is that many of the GIS information services that are provided to businesses, we're talking about businesses here, more than consumers. Is ultimately informed by census data or in some cases, other sources of government statistics, is that right?

 

 

>> Feng Chi: Yeah, exactly, and also for those private data sources, because of their smaller sample size, they sometimes have biases. They're good at tracking changes over time, but they really rely on the government official data as a benchmark to correct for these biases and provide a full picture.

>> Steven Davis: Let me take the skeptical stance.

 

Suppose I still say, okay, but look, there are many sources of information. Even if decades ago we couldn't really get this kind of information except from government sources. There's been a proliferation of private sources of information that many private parties put that out there for their own purposes.

 

Either they wanna sell it or they wanna do brand development or thought leadership. There are many such examples of private sources of information that's also useful for businesses. So suppose I'm a skeptic. I'm on the DOGE team and let's be cutting all this government spending, we don't really need it.

 

So one of the cool thing about your research, okay, which really makes it kind of, which caught my attention. Is you have a clever but transparent way to demonstrate that on top of whatever other information might be out there in the ecosphere. You can show us that the quality of business decision making around things like site selection that then affects things like business survival is very much influenced by the availability of the decennial census data.

 

Okay, so tell us, how is it you demonstrated that on top of all the other information sources out there, there really is something special and additive about the decennial census data? What's your basic idea? How did you approach this issue?

>> Feng Chi: Yeah, so the key idea is actually about timing.

 

The decennial census, by definition, takes place every ten years. As you mentioned in the intro, this is mandated by the Constitution. And the original purpose is to apportion seats in Congress based on population counts. Even though businesses benefit from census data, this data is collected based on an exogenous and predetermined schedule.

 

It just happens every ten years. So when the data is freshly collected, we get a very accurate snapshot of local demographics. But then things change over time. People move around and neighborhoods evolve. These changes don't happen overnight, but over a course of ten years, things could really change.

 

So what used to be accurate data that reflects the current condition of the neighborhood becomes outdated. And then if you enter the market later and had to rely on stale information, you're more likely to make bad decisions and will see consequences.

>> Steven Davis: Okay, so just to be very concrete about it, the decennial census is done in years that end in 10.

 

So there was a decennial household census in 2010. And what you're telling me is that the information that comes out of that census is much fresher in 2012 than it is, say, in 2018. So if you're a business in the retail sector and opening a restaurant, personal services, you've got better, fresher information about local demographic and socioeconomic characteristics of people in an area in 2012 than in 2018.

 

So that's the key observation. And then you're going to compare, basically, well, what happens to the businesses? Tell me, if I got it wrong, what happens to the businesses that entered in 2012? How do they do in the succeeding years compared to the businesses that enter in 2018?

 

That's the basic idea, but then you're gonna compare. Compare businesses where this kind of detailed local information is really important versus those where it's less important. So we'll get into what the logic is there. But first, if I got the basic idea right?

>> Feng Chi: Yeah, that's exactly right.

 

Because the census takes a regular interval based on the schedule. If we think of when census was first released as the origin point, and at this point, data is more fresh and business can make better decisions and new business are less likely to fail. And if we stack them based on these entry cohorts when they enter the market, we'll see failure rate gradually goes up over time as business now have to rely on more outdated data.

 

 

>> Steven Davis: And here, correct me if I'm wrong, we're specifically talking about the types of businesses where this local information is especially important. So we're not talking so much about a manufacturing plant. We're talking about things like retail services, accommodation and food services where you really need to know your local area.

 

That's where these effects show up most prominently. Is that right?

>> Feng Chi: Yeah, absolutely. So my sample is based on businesses in the retail, and accommodation and food services sectors. One reason I choose those sectors is because they really rely on local information. But they also have this other advantage that allows me to study to isolate the fact of information on entry decisions.

 

So, if we're looking at manufacturing facilities, success and failure also depends on macroeconomic conditions. If things change over time, we're not sure if this is really due to information difference or something else.

>> Steven Davis: Okay, so that's a great point. If you're talking about an automobile company, its fortunes are not rising and falling with the fortunes of the local community.

 

They're rising and falling with national economic fortunes and even global economic fortunes, right?

>> Feng Chi: Yeah.

>> Steven Davis: Whereas if you're looking at a food service outlet, it really depends on the local community. And there are many, many local communities across the United States. So just in a statistical sense, you get a lot more power, a lot more informativeness about the question at hand by focusing on retail trade and accommodation and food services.

 

That's your thinking, right?

>> Feng Chi: Mm-hm.

>> Steven Davis: Okay, so give us a headline number. We won't go into too many numbers. But if I'm listening to this and I wanna get one, we haven't talked about how big these effects are, how small they are. If I want one kind of takeaway headline result about how big this effect is from information staleness increasing over time following a census, what would it be?

 

What's the one number you want people to take away?

>> Feng Chi: Yeah, so as you move away from when census data is still fresh, failure rate will go up over time each year. And then over the time interval of 10 years between two decennial censuses, it increases by 16 percentage points.

 

So we're comparing firms that are exposed to the most fresh data and firms that had the worst data. The difference in their failure rate is 16 percentage points. And just to give a sense of the magnitude, so the baseline failure rate in my sample, which is on average, 50% of the retail establishments in my sample fail within the first five years.

 

So that's the benchmark. And if you think about the 16 percentage points increase over time, the difference between those with worse information and those with best information is about 32% of the baseline.

>> Steven Davis: Okay, and so here, again, we're talking about businesses in retail trade and accommodation and food services, where there's lots of business entry, there's lots of failure all the time.

 

That's the character of these sectors. And so when we go from businesses that entered with the freshest data right after a census to businesses with the most stale data, the failure rate goes up by about a third. So that's a big effect. And so that's back to the point we were making earlier.

 

This is despite all of the other information sources businesses might have from personal observation, from private sector sources as well. So what you're finding is evidence that the decennial census data is a big enough impact on the information environment that it has a very sizable effect on business success as measured by failure rates.

 

And failure is a pretty coarse measure of lack of success, but it's a good one to use for this purpose cuz it's easy to observe, easy to measure with confidence, right?

>> Feng Chi: Yeah, so one thing I forgot to mention is, previously, when I was talking about the setting of this paper, I looked at those retail establishments.

 

Because you can compare those who just entered the market because they need information to make this entry decision and pick the correct market to enter, versus the ones who are already operating in this space. So if there's fluctuations in macroeconomic conditions, it will affect both the new entrants and existing establishments.

 

But then I can use the existing establishments as a control group to take out the things back.

>> Steven Davis: Okay, so now this is an important point. Statistically, we're getting a little bit deeper into how the analysis is actually carried out. So maybe I misled people a bit. You're not just comparing failure rates of the ones with the freshest data to the most stale data, but in each case, you're doing what economists call a difference in difference.

 

You've got the failure rate of the new businesses in a given area in a given industry compared to the failure rate of older established businesses in the same area in the same sectors. And so that difference between new and old businesses when information is stale is fresh. You're comparing that to that same kind of difference when the information is fresh.

 

And that's, go ahead.

>> Feng Chi: Yeah, so this really allows me to contribute this difference to how firms make entry decisions based on this public data. So, those establishments are not fundamentally operating worse than others. It's really because they pick the wrong markets. It's a bad market fit and it's because it's based on this bad information.

 

If they had picked a better market based on better information, maybe they would have survived.

>> Steven Davis: That's a good point. So, maybe it's obvious, but maybe it's not. You're not saying that the people who enter the market by opening new businesses when information is stale are somehow worse business people.

 

They're just operating in an environment with worse information, and that's what leads to their higher failure rates. It's the worst information, so this is drawing out the economic value, the social value, the commercial value of the decennial census data value in the private sector. As you pointed out earlier, and as I said at the outset, the decennial census is a constitutional requirement originally created for public purposes, but it has private value.

 

And you're using the every 10 years feature of it to, to draw out what that private value is in one context. Now, I'm going to let me continue to play the role of the skeptic and say, okay, well, yeah, that sounds sort of plausible. You've got something there, but what if there's just something else going on?

 

I don't know what it might be that in say, 2018, it was just for some other reason, not such a great environment for new business entry as 2012. And your results are really an artifact of something else we haven't controlled for, I might not even know what it is.

 

So you do a bunch of robustness checks, you might call them, or additional analyses to try to say, no, that's really not what's going on. Describe some of those.

>> Feng Chi: Yeah, so that's why I don't want to just rely on the timing of the census. I also want to do some cross sectional checks to confirm that the channel is really through this information.

 

So one thing I did is I compare areas that has experienced large changes in demographics over the decade versus the ones that had little changes. And for those that had little changes in demographics over time, old data is not really an issue. So new data would have been the same as old data anyways.

 

And that's where I find no effect on failure rate in those areas. But in areas with large changes in demographics, those are areas where the current market condition has significantly deviated from what the data says. So when firms enter these markets, they don't get what they expected. They might end up in the wrong markets and they fail more.

 

 

>> Steven Davis: Okay, so, yeah, so that's a nice test. Let me explain it in a different way. So there's a decennial census in 2010. There's another one in 2020. By comparing the numbers in 2010-2020, we can see which local communities really had stale and misleading data by 2018. Because by 2020, we actually know what happened with high confidence when the next decennial census comes out.

 

What you're telling me is that when I go back then and make use of what I've learned by 2020 about which communities had really experienced big changes after 2010, and which had not. It's the ones that really had the big changes where you see the big increases in failure rates among new businesses in the 2018 compared to, say, 2012.

 

 

>> Feng Chi: Yeah, exactly. You put it in a nicer way than what I had.

>> Steven Davis: No, but that's a very nice test. And it's very hard to see how that kind of result could be explained away by something else that just happened to make 2018 different than 2012 for new versus old businesses, right?

 

 

>> Feng Chi: Yeah, cuz the key logic of the story goes through whether those areas had experienced large changes and still data cannot capture this change in areas that with lots of changes, that's exactly where I see the results.

>> Steven Davis: Right, and you had some other things you did that kind of confirmed your central interpretation of the finding.

 

So as I recall, you distinguished between businesses where high quality, highly localized information was really important. Versus businesses where it was less important to get high quality, timely information about the local neighborhood. If I got this right and can you explain that a bit?

>> Feng Chi: Yeah, so I also look at, so these are businesses in retail and food and accommodation services, but within this big sector, there are different retail businesses.

 

And my results are strongest for grocery stores and restaurants. So these businesses offer really localized products and services that people get almost on daily basis. They're unlikely to travel very far frequently to get these. So it's very important for these businesses to be located where the demand is strongest.

 

So those businesses have a stronger, it's more important for those businesses to get localized information. And that's where I see the strongest effect.

>> Steven Davis: So if I were to just take a counterpoint, you've got the restaurants and grocery stores on the one side. But then a different kind of business might be an appliance retailer or an automobile dealer that you don't buy a car every day.

 

And so when you do buy, and it's also an expensive item, people are gonna look over a wide area for potential places to buy a car. And so it's less important that the car dealer have a really precise read on the demographic, say within one mile of the car dealership than it is for a grocery store or casual dining restaurant.

 

That's the idea, right?

>> Feng Chi: Yeah, so, absolutely. So those subcategories that sell cars, furnitures, you mentioned, they have a much larger trade area, which means that customers come from different places. They're willing to travel further because it's a bigger ticket item, durable, good, they could use for many years.

 

So they will make an effort to pick and choose the better deal. So the precise location of where your dealership is is not as important as like say restaurants and grocery stores. And that's where I see they the outdated census data does not have much of an impact on these sub industries.

 

 

>> Steven Davis: Right, okay, so let's summarize what I take away at least from your paper, and you can tell me if I've missed something central. Headline result is when we go from the freshest data shortly after a decennial census to the most stale data right before the next decennial census.

 

That that leads to an increase in business failure rates in the retail and trade sector, food and accommodation that's the sectors we're looking at. That raises the business failure rate of about one-third. Okay, compared to baseline. So big increase in failure rates. And your interpretation is that it's because of the deterioration in the quality of information that they have to make use of to decide where to locate their business, what products and services to offer.

 

That's the headline result. And then you had a few other results that kind of confirm your interpretation that that's really what's happening. When you looked at across local areas that either changed a lot or a little between decennial census. And then when you looked across different kinds of businesses that are either highly sensitive or not so sensitive.

 

To precise, high quality information about the demographic and socioeconomic mix of people in a small geographic area around the business location. So it sounds you've got pretty compelling evidence that the decennial census really does add a lot of economic value, of course, for the businesses, but not just for the businesses, for everyone else too.

 

Because the consumers wanna have businesses that are well suited to their own tastes and interests and what they want to purchase and price the way that's suitable for them. So even though we've been talking about businesses, we shouldn't lose sight of the fact that this information has value for consumers as well as businesses and for workers.

 

Cuz better information means fewer business failures, fewer lost jobs. That's the story, right? Is there any else big picture that I missed or that we've missed from your study for the Decennial census, go ahead.

>> Feng Chi: I guess I have another set of results where I show that the outdated census data really has a larger impact on small business compared with large businesses.

 

So, large businesses often have access to other sources of information that the small ones cannot afford to get. So if you're a large chain, you already have existing customers. So each time when you go up to a store and they ask for your zip code, this also gives them information about where are the next markets they should expand into.

 

They can also afford to conduct large scale consumer surveys and also get alternate data from other sources. But if you're a small firm or a small entrepreneur, what you rely on is the public data sources. So you're at a disadvantage when the public data does not reflect the current market condition.

 

 

>> Steven Davis: So this is a great point. So this information is more valuable on average to smaller businesses than larger businesses. Because larger businesses have other sources of information that either are unavailable to the small businesses or are too costly for the small businesses to acquire. And it's interesting because rightly or wrongly, many politicians want to promote small businesses.

 

And they do it through a whole variety of schemes, some of which have quite doubtful value or even counterproductive in my view. But here you're pointing out that there is something that the government does routinely, as a matter of course. Which is generate statistics about local economies that turns out to be especially valuable for small businesses and helpful in their success.

 

So that's quite an interesting and important observation. So if you're a politician listening to this and you want to think about how can I support small businesses in my communities? High quality statistical information is part of the answer.

>> Feng Chi: Yeah, cuz I think data inherently is a public good.

 

Just because you use the data does not preclude me from using the data. And this public provision of high quality data is a good utility that could benefit everyone in the community. Just philosophically, I believe it's better than policies that directly say if you want to subsidize those smaller retailers or intervene in terms of like, competition practices of those, like large chains.

 

You're just giving good information to everyone so that they can compete fairly in this market and the market will do its own work and let the best, most efficient ones stand up.

>> Steven Davis: Right, well said, so let's step beyond your paper now. I wanna broaden the discussion for the last part of our conversation just to explain more broadly the sources of private value that come from government statistics.

 

And why it's hard to fully reproduce that from any private sector source. And I invite you to add your own points here, but I have a few things, I want to say and get on the table because they're not all, some of them are non obvious. First, the US statistical agencies have better sample frames.

 

And here I'm using jargon, they have a better sense of the population from which you might want to sample from than do private parties. The sample frame in the case of the decennial census is all the household and dwelling locations for persons throughout the United States. US statistical agencies, the census maintains that frame on a regular basis from which they can then draw random samples not just for the census now.

 

But for many other surveys that the government conducts, including one of the most important being the current population survey. Which is the source of unemployment statistics that we read about in the newspaper every month. And the government also has sample frames for businesses that feed into again reports that the government puts out all the time, those frames.

 

It's very hard for a private sector party. It would be extremely expensive to reproduce a frame which is as good as the ones in the government sector. So what that means, and you alluded to some of this earlier in your remarks, all these private sector surveys that are done, which there are thousands and tens of thousands, I do some myself.

 

And they're very useful, they're very valuable. But they often benchmark themselves against government sources to get a more accurate statistical portrait. So even all of the private sector survey activity that's done, which now these days may be larger in scale than public sector survey activities. It's often, usually, I would say benchmarked against public sector sources to make the private sector information more valuable, more accurate, more reliable.

 

So that's kind of point one, and that's really important. So private sector data is more valuable because it can be benchmarked to public sector sources like the decennial census but also to other public sector sources.

>> Feng Chi: Yeah, I think there's really complementarity between the public say decennial census data that provides a benchmark for the private data sources to correct their biases and provide more timely updates.

 

I think the Census Bureau actually has tried to look at some of those private data sources to bringing information and reduce the cost of conducting large scale skilled service. So what they found is those private source data are pretty good at tracking changes over time but they really need the benchmark to provide a better picture of what's actually going on.

 

So these public and private sources are complementary to each other.

>> Steven Davis: They're complements. That's right they're complements. Each is made more valuable by the other. So let me make a different point again this is directed towards people who are skeptical about the value of government statistics. There's abundant evidence that government statistics when they are released move financial markets, stock markets, sometimes interest rates and often sharply.

 

So obviously financial markets think there's real economic value telling us about future economic outlooks that get, that affect things like stock prices, and I'll give one example. It's perhaps the best known example of this. Every month, the Bureau of Labor Statistics issues what it calls the Employment Situation Summary.

 

It's often colloquially referred to as the jobs report. And that report comes from a survey of businesses and how the current employment statistics and another survey of households which we already talked about, the Current Population Survey. This Employment Situation Summary is perhaps the most closely watched statistical report around the world that gets issued every month.

 

So you'll see newspaper articles about it every month. Central banks around the world, including most obviously the US Central bank, the Fed, are paying attention to this. So it clearly matters, it moves markets. So that's kind of a second point. And then the government data. Government statistics are the source of data that we use all the time about wage and price movements.

 

And they're essential for tracking inflation and living costs and compensation trends. These data feed into cost of living adjustments for Social Security benefit payments, disability insurance payments, federal retirement pensions. So again, were really dependent in many ways on statistics that are generated by the public sector. So I could go on in this vein, but I'm just trying to make the case that beyond the Census Bureau Statistics itself, which are mandated by the US Constitution, as we discussed.

 

The US Government and state and local governments provide other statistics that like the decennial census, have value in the public sector and in the private sector. And I don't think we wanna lose sight of that when we're thinking about how we wanna spend taxpayer funds.

>> Feng Chi: Yes, I think in general we're really entering into an era of data driven decision-making.

 

This applies to companies, to the public sector, and to every individuals. So at the end of the day, you really want to provide good data and to inform people's decisions.

>> Steven Davis: Okay, so I think that's a great note to end it on. Good data leads to better decisions, better economic outcomes.

 

The sources of the good data come from both the private sector and the public sector, and the two are complementary. Public sector statistics make private sector data more valuable and vice versa.

>> Feng Chi: So what I wanted to mention is what my paper could answer is what happens when there's no freshly updated data in between the censuses.

 

But what it cannot say is what if there's no decennial census at all? So if we don't even have this benchmark, this could even affect larger companies and those who have private data sources.

>> Steven Davis: Right.

>> Feng Chi: So the impact could even be larger than what I document here.

 

 

>> Steven Davis: May be large, yes. So we don't know exactly how bad things would might get if we had no decennial census or other sources of government statistics about socioeconomic and demographic outcomes. But your paper makes a pretty good case that it would lead to a lower quality information environment.

 

Which would mean worse business decisions that would be bad for businesses, that would also be bad for consumers and bad for overall economic performance, okay? Thanks so much, Feng, and good luck at the University of Illinois when you start there, I guess in the fall.

>> Feng Chi: Thanks again for having me.

 

It's been a great pleasure.

>> Steven Davis: Okay, take care. Bye bye.