This is Planet Money from NPR. Amy O'Hara is a data person.
Trained as an economist, worked at the U.S.
Census Bureau, now works at something called the Massive Data Institute at Georgetown.
And that's where she was last Friday, getting ready to go out of town to Nashville for a big statistics conference.
It would have been a pretty chill Friday before travel.
You know what I mean? But it wasn't. It wasn't because in the middle of the day, Amy comes out of her office to meet one of her colleagues.
He's coming to pick up some stickers that they're planning to bring to that conference.
Stickers all about how important good data is.
But when Amy comes out of the office, she realizes that her colleague does not look happy.
He's like, have you heard? Wow, did you hear about BLS?
I said, no, what about BLS? BLS, the Bureau of Labor Statistics.
As you may have heard, they released their monthly jobs report last Friday, and it was a bad one.
The economy is adding fewer jobs than people expected.
And the BLS also released their revisions to the previous two months of jobs data.
And those two months were worse than we had thought.
Revisions always happen, and these revisions were big, but within a normal range.
Basically, the report said that the labor market is not doing so great.
President Donald Trump, upon getting this bad news, fired the head of the Bureau of Labor Statistics, a career statistician named Erica McInturfer.
Amy and her colleague both know Erica, and they're both shocked.
Just literally standing there in silence for, I don't know, 20 or 30 seconds, letting it sink in, because something had dramatically shifted.
You know, it was that the norms that statistical information could be produced without political interference changed.
Those norms were shattered. The BLS is responsible for all kinds of important government statistics.
The unemployment rate, the inflation rate.
And the reason this firing is such a big deal to Amy and her colleague, and a lot of people, is having solid, nonpartisan economic statistics from the government.
That is important. Local governments rely on that kind of information to know what's going on in their area.
Businesses use it to plan for the future.
Decision makers all over the economy rely on these numbers to be rock solid.
And now, the head of the BLS had been fired.
This was a direct response to numbers that were politically inconvenient.
And that raises all kinds of worries. that the government could try to pressure these independent statistical agencies to suppress information or even to put out something that's manipulated or false.
Amy is worried about that. And she's worried about smaller, subtler changes.
The BLS and agencies like it have already been losing staff under Trump.
There's a worry that could make it harder for those agencies to do their job well.
And if the government starts firing more of these staffers, perhaps for political reasons, that is fewer people to uphold norms and processes.
Maybe the statistics will slowly get worse.
What I think is scary is as a consumer on the outside, what I notice is So that sort of change detection is really tricky, that if the number looks a little different, or even if the number looks exactly the same, how I consume that, how I interpret it, has always relied on the full trust that the feds were doing a good job putting out numbers.
And anything that deviates from that, to me, is scary.
At the big statistics conference this past week in Nashville, these concerns were front and center.
There is sadness. There is shock. You know, there's this... this new uneasiness, you know, a wariness about what we have to be looking at, what could happen next.
What could happen next? Amy told me she is not jumping to the worst-case scenario.
And neither are we. There are some reasons to believe that U.S. data collection might be okay.
The systems are built to prevent political tampering.
They're decentralized. They involve many people.
The commissioner doesn't even see the jobs numbers until very late in the process.
And there are many BLS employees who could blow the whistle if they saw anything amiss.
But political data manipulation, it has happened.
It has gotten bad. And it is worth remembering exactly why manipulating economic statistics is really not a good idea.
Hello and welcome to Planet Money. I'm Sallie Helm.
And I'm Mary Childs. Today on the show, two historical, statistical, cautionary tales about what happens when politicians get tempted to cook the books.
And once you've cooked them, can they get uncooked?
This message comes from NPR sponsor, U.S.
Bank. With U.S. Bank Business Essentials, you get more than just a bank.
You get a dedicated partner that provides you a powerful combo of checking and card payment processing with quick access to the money you've earned, proving that there is nothing as powerful as the power of us.
Visit usbank.com today to learn more. Member FDIC.
Copyright 2025 U.S. Bank. Today, we offer you two historical statistical cautionary tales.
Tale number one, how to cook the books. We begin in Argentina, 2006.
Inflation has been high, and no one likes high inflation.
I remember when inflation was around 9% in the U.S. in mid-2022.
I feel like I still have not emotionally recovered from those grocery bills.
Honestly, same. Argentina in the mid-2000s had it even worse.
And it got to the point where it was reaching two digits, about 10%.
High. Yes, high. You know, Argentina has gone through many inflation crises in the past.
So it was obviously a politically sensitive issue.
Economist Alberto Cavallo grew up in Argentina, and he has studied this period closely.
Argentina is indeed no stranger to high inflation, even higher than 10%.
But still, it hurts. And around 2006, an election is coming up, and the government of Argentina seems to have decided that instead of going through something painful like higher interest rates…
Maybe they could do something simpler. Just make that high inflation number become a different lower inflation number.
I think they basically said, OK, we don't like the number.
Who is putting together this number and how is it being put together?
In Argentina, the number is put together by an agency called INDEC, the main statistical agency in the country.
Quick refresher here. Inflation is calculated based on the prices of a particular basket of consumer goods and services.
A group of things that any given person in the economy is likely to be buying.
Things like food, clothes, furniture. INDEC and agencies like it, including the Bureau of Labor Statistics here in the United States, they will track the prices of those goods over time.
Boom. Inflationary. And to do this, these agencies send people out to real stores to ask, what is the price of the shoes you're selling?
Or the lettuce? Or the microwaves? Indec had a particular group of stores that they would return to over and over.
And the Argentinian government came up with an idea.
So the first thing they did is they went to a statistical office and said, can you tell me the location and who are these stores that you go to and you collect data from?
And why were they asking that question? They were essentially, I think, at the beginning trying to figure out if they could go, possibly pressure these stores, see if they could do something in that process of data collection that would help them bring that number down.
That ever-important inflation number. This is idea number one for messing with the inflation rate.
Alberto says the plan seems to be this. If the government could find out where these statisticians got their data, maybe they could influence that data.
Lean on particular store owners to keep their particular lettuce or shoe or microwave prices low.
Then the inflation number would look lower, even though inflation rates had not actually changed.
It's a sneaky way to cook the books, by changing the ingredients.
But those statisticians said no. They wouldn't reveal the stores.
This first attempt to bring the inflation number down in Argentina failed.
So the government had to try a different tack, a second way to cook the books.
You could tweak the statistical recipe for inflation.
They started saying, well, can you make methodological changes?
And some of these pressures started materializing that way.
Methodological changes. That's so interesting because it speaks to the fact that there is a little bit of room in the way these numbers are calculated.
Can you give me any more specifics? How did they try to mess in that space that they had?
You're absolutely right. Every one of us experiences a different inflation rate because we buy different things.
So the inflation rate is meant to be, you know, an average that represents a typical consumer in the economy.
So one of the things they can change is the actual percentages that are assigned to each of the different categories.
Like, I don't know, if health care costs are going up a lot, you could change how much weight you're putting on the health care part of the basket.
Put more weight on, like, furniture. Or at least theoretically, you could change the weighting.
Frankly, it would be pretty obvious. Because statistical agencies report the weights they're using and any changes in those weights would be immediately flagged and people would question why is this happening.
The government suggested other, subtler methodological changes, but the statisticians refused to play along.
This idea for messing with the inflation number also didn't work.
After about a year of this pressure in 2007, the person in charge of calculating the inflation number in Argentina, she got fired.
So did some other statisticians. The government said, look, we've gone there.
We've discovered there are all these methodological issues.
We're going to make this a better statistical agency.
We're just going to fire these people because they didn't do the right job.
And he says, you can imagine at first some people watching all this are like, OK, fine.
Who knows? Maybe they were bad at their jobs.
But then some of these fired statisticians start speaking out, telling these stories about the pressure that the government had been applying.
That's when people like Alberto learned that all this had been going on.
And Alberto says weird things started to happen, like some data went missing.
There were some pieces of information that started disappearing from the website of the statistical agency.
They stopped publishing some more detailed indices, which started to raise suspicions.
But again, nobody really knew exactly what was going on.
After this point, the inflation numbers coming out of Argentina seem too low.
Lower than they should be. People are pretty sure the government is now cooking the books.
So economists decide to try to figure out the real inflation number.
Bye. knocking on doors themselves. Economists were sending people to the stores to build these small baskets of goods and try to measure inflation.
But the government would reply that, you know, these are very small samples taken in very specific places.
The government did more than try to discredit the numbers.
They threatened statisticians and economists with lawsuits that could land them in jail.
They issued huge fines. Alberto was in the U.S. at the time.
He was studying at Harvard. He was safe from this retaliation.
And he wanted to help. So I had this idea.
I said, I can actually try to use the data that is available online and build this index using data collected from the web.
Alberto decides to try his hand at calculating the true inflation number using this new data source, these online prices.
Because the government doesn't have a monopoly on asking store owners, how much does this lettuce cost?
The thing they do have is scale. So many employees.
It was hard for a rogue economist to match a huge agency like Indec.
Alberto's solution is to create an online index.
He's able to scrape a ton of prices from major online grocery stores in Argentina.
That helps solve the too-small-basket problem.
And he uses those prices to calculate inflation, specifically something called the supermarket index, which is similar to something that the government publishes.
And he did this not just in Argentina, but also in Brazil, Chile, Colombia, and Venezuela.
And he found that in those four other countries, his online price index matched up well with the official numbers, but not in Argentina.
I was finding levels of inflation that was two or three times higher than what the government was showing.
And my feeling was if the government wants to criticize this, they actually have to tell me that the price of milk is not what I am showing on this website.
So the government actually ignored me, but I started getting a lot of emails saying, from users of the website that were very thankful.
Because if you think about it, inflation is a tremendously important indicator, not just for the government, not just for investors, but consumers have to make decisions.
Businesses have to make decisions that are based on this.
Yeah, people need to know how much inflation to expect so that they can plan a big purchase, like buy a car now or wait till next year.
And people in Argentina had just fully stopped trusting the government inflation number.
Which, fair, the number was a lie. And Alberto really wanted to figure out how was the government even calculating that number?
Like they must have some formula behind the scenes.
He tried and tried to take his price numbers and get them to give him the government's inflation number to figure out how precisely the books were being cooked.
So I tried lots of things. I could borrow your world technical details, but I could never get the right number.
Really? And then I realized there was an empirical regularity.
If you looked at the levels of inflation... they were always about three times higher than what the government reported.
So if you divided the real inflation rate by three, you got the actual number that the government was reporting.
Divide by three. That seemed to be basically the algorithm that they were using.
Which sounds funny. At the beginning, you would think, if they're going to lie, why use such a simple algorithm?
And there are two things I suspect happened there.
One is that You've never lied in your life, and I'm sure all of us have.
Never, never. No, exactly. But if you have to lie, you also have to be dynamically consistent.
So what do I mean by that? He means that the inflation rate had to move up and down dynamically.
The government was not simply always announcing that the inflation rate was falling.
That wouldn't have been plausible. The number did go up and down in a way that made some sense, given overall booms and busts.
Divide by three is just a simple way to achieve that, to keep things roughly moving in the right direction.
Just make the overall number lower. But there's another thing that Alberto suspects might have been going on here, and it has more to do with the behind-the-scenes factors at an agency like INDEC.
If the government wanted to lie, they basically had two ways of doing this.
They could replace everyone in the statistical agency that was collecting all these prices and get them to lie as well.
Very hard to do. You basically have to replace hundreds of people.
The other thing you could do is you replace the people at the top that aggregate the data.
So everybody sees a little piece of the data, but not the aggregate numbers. it would be simpler to replace or influence just some smaller group of statisticians at the top so that all the other statisticians can do what they normally do.
And it's only at some point later down the line when all the pieces get put together that someone somehow messes with the number.
However they did it, the government lied about the inflation rate for years.
And everyone knew it. Alberto actually later studied how people in Argentina reacted to these manipulated statistics.
Because he was curious. Did people just ignore them?
Or did they try to interpret them? Read the tea leaves.
And he found that people were listening to what the government said.
When the government reported higher inflation, people were like, yeah, must be even higher.
They had an intuitive sense of the real number.
But when there was good news, they didn't trust it.
People suspected the worst because they thought they could not trust anything that the government was actually doing.
People had just lost so much trust. They started assuming that things were bad, even if things actually had gotten a little better.
Which goes to show, simply making the number look better does not actually change people's minds.
It might even backfire. Because we know what the prices are.
Not just from reading the statistics, but from living our lives.
What finally ended this was an election.
The lie went on from about 2007 until the government turned over in 2015.
It took the new people a while to get things up and running again to start publishing the true inflation numbers.
So for about six months, Argentina had no inflation index.
And when it came back, it actually started reflecting very close to what we were collecting online.
So that's when I knew things had gone back to normal.
I asked Alberto about what's happening right now in the U.S. with the BLS commissioner being fired.
He said he thinks the firing is really bad for the credibility of the institution and potentially for the quality of statistics in the long run.
Of course, a lot depends on who gets appointed next.
But in general, he said the U.S. is starting from a really good place with a lot of solid statisticians.
And when it comes to the inflation number in particular, he's not that worried.
There are so many private indices now, it would be really hard to mess around.
I would be more concerned if we are talking, for example, about the jobs report, because that's the kind of thing that the private sector cannot easily replicate because the statistical agency runs these massive surveys all over the U.S., which are very hard for the private sector to replicate.
But he says he'll be watching for any signs that the books are starting to get even a little bit cooked.
After the break, another country gets a crash course in book un-cooking.
We look at statistical historical cautionary tale number two, what happened in Greece.
This message comes from Osea, who believe the key to good skin is in the ocean.
Osea's Undaria Algae Body Oil harnesses seaweed for silky, soft, and glowing skin.
Get 10% off your first order with code SEAWEED10 at oseamalibu.com.
Alright, time for cautionary tale number two.
How to uncook the books. When you're thinking about uncooking the books, there is someone very obvious to call.
Hi, I'm George Papakonstantinou. I'm a professor at the European University Institute and a former minister of finance in Greece.
So you're an economist. What do you teach?
I teach economics. I teach macroeconomics and I teach crisis because I've been through a number.
The most salient crisis for George and for us today is the Greek financial crisis.
We extracted this one little story from that time and asked him to relive it for us.
For George and for Greece, it started in 2009.
Greece had an election in October that year for prime minister, and George's party, the center-left party, wins.
We won the elections with a huge landslide, and I got a new job.
His new job was finance minister. So it's fall in Athens.
Economies around the world were kind of in shambles in the aftermath of the global financial crisis.
But in Greece, George remembers a lot of optimism about the new administration.
People were feeling good. October is the beginning of the season, so people are back from vacation.
They are fairly relaxed. There was, on our side, a lot of jubilation and huge expectations.
So George gets his new office, and he sits down, and he's like, OK, where are we at?
Let me pull up some of these documents. We kind of knew things were bad.
We just didn't know how bad things were.
One of the first things that happened was the person responsible within the ministry for the for the accounts, comes into my office and says, look, things are actually quite bad, much worse than the official announcements.
Oh, really? They just confessed right away?
Yeah, because that person was doing their job right, and they were preparing monthly statements, which they were giving to the minister, and the minister was just putting them in his drawer and not announcing them.
So over the summer, there was no published data.
No published data. The government just wasn't saying anything about the deficit.
In Argentina, the magic number that everybody cared about was inflation.
But for Greece, that magic number was the budget deficit.
How much money the government was spending versus how much it had coming in the door.
And at the start of the year, the prior administration had projected that the budget deficit was going to be 3.7% of GDP.
But then after several months, they'd started saying, you know what, it's actually looking more like 6%.
And this number is an important barometer of a country's fiscal health.
The lower, the better. If it's low, that means the government hasn't borrowed too much relative to its economy.
Everyone is happy to lend to this country.
If it's high, there's too much debt and not enough economy.
And people get worried. And part of the deal of being in the European Union, as Greece is, is you're supposed to keep your deficit under 3% of GDP.
And the way to calculate that ratio, like, those inputs are generally pretty straightforward.
There are rules. But the Greek government was being pretty creative with how to follow them.
Like, how much of something should count and when do you count it?
They were basically taking a lot of liberties with those gray zones and hiding stuff that should have been in there.
So, if it's worse than 6%, George is like, how far off are we here?
This is step one to uncooking the books.
Once you find out that the books have been cooked, you have to figure out how much cooked.
So George starts calling people. He calls the people in charge of taxes, the people in charge of spending.
He calls Greece's central bank, the Bank of Greece.
And he starts getting answers. I mean, they were underestimating expenditures and overestimating revenues.
Hence, the deficit projection for the end of the year looked to be 6% of GDP, whereas in fact, we were October.
And the Bank of Greece data were telling us that we're already at 10%, going up 1% every month.
So we're going to hit at least 12%. So by the end of the year, the estimated deficit would be at least 12 percent, twice what the government had been saying.
George says he was like, how does a mistake like this happen?
So I called in the head of the statistics office and I said, what's going on?
George is like, we got three numbers here, 6, 10 and 12.
And yours is the lowest by a fair amount.
So which number is real? So he said, well, you know, all we do is project, right?
So basically, I fired him on the spot because I felt quite clear that he was hiding the truth.
Basically, he was feeding the previous minister the numbers that the minister was asking him to give.
For the record, that guy has since passed away.
But back in 2010, he did deny wrongdoing and he told The New York Times that he resigned.
So step one is done. George now knows the numbers are off, maybe double off.
And here is his next problem in the process of uncooking the books.
There is now a gap between what he knows and what the public knows.
The last they heard, the deficit was 6%.
And they're all still out there feeling great.
So at some point, he's going to have to burst their bubble.
You have to sort of calibrate how you break the bad news because you don't want to completely dash people's hope and expectations, but you also don't want to mislead them.
George and the administration, they need people to trust them.
They need to somehow reestablish credibility. so they can fix the problem and move forward.
George says the next step is obvious. They should come clean.
But he says that was not a universal opinion.
Because some people actually at the time said to us, why don't you continue sort of hiding it a little bit, you know?
We said... This thing is, we owe this money, right?
And if you keep lying, the lie gets bigger and bigger.
You cannot do it. We're not elected to continue lying.
So we have to come clean. This is step two.
Confess. Unlie. To do this, George says he calls in the governor of the central bank.
And they go stand in front of a bunch of journalists.
We both come out and he says, you know, it's going to go above 12%.
So that's a significant moment because that's the first time when it's out in the public.
So now everybody knows. The budget deficit is not 6%.
It's closer to 12%. Sorry. That is step two accomplished.
But George doesn't get much credit for coming clean.
His message was not received with a thank you for letting us know.
Because we still have a problem. Reality.
Here comes step three, the actual manual un-cooking.
They have to figure out reality, what the deficit number actually is.
Not the 12% estimate, but the real exact deficit for 2009.
And for that matter, the past few years too.
So for this, they have to tally up all the money that literally came in the door and all the money that literally went out the door, which is a lot of work.
I have a couple of weeks left which will forever stay in my mind, where we have basically morning to evening meetings with a huge cast of characters coming in and out.
So we can actually try to nail down the numbers.
So we've got the people from the revenues and the people from the expenditures and the people from the statistical office and the people from the Bank of Greece.
And at the end of every day, we have a new number.
And I go, is this it, guys? And they say, yes.
And then there's always this little voice from the back of the room, a guy raises his hand and says, actually, minister, there's one more thing.
And then there's this additional item, which is hundreds of millions of dollars, which has not been counted and needs to be because it has to be, obviously, right?
And then the number keeps going up. George tells us one example.
There was some public corporation that the Greek government had sold, but when they sold it, they for some reason kept the pension obligations.
George says the government was dutifully paying those obligations to retirees, but just not including that outgoing money in its deficit calculation. which means that every year there was something like 700, 800 million, which were not written in the accounts, even though we were actually paying it.
It was a clear attempt to make sure that the deficit did not show as big as it really was.
Yeah, George says somehow those numbers, which should have been included, were getting lost on the way to the deficit calculation.
This all takes over a year. It is so hard to get the information.
People are dragging their feet. They hire a new head statistician.
The European Statistical Agency sends in their people.
And eventually, the weeks of morning to evening meetings are finally done.
And they have finally heard from all the guys in all the backs of all the rooms.
Once we counted everything, the 2009 deficit ended up being 15% of GDP.
So from 6% to 15%. That's 9 percentage points of GDP missing, which is what made the financial markets completely freak out.
The financial markets completely freaked out.
And this is step four. You know how cooked you truly are.
You've confessed. And now it's consequence time.
And there are many consequences. They show up first in financial markets, the bond markets.
Investors around the world who had happily lent Greece money in the past were now demanding that Greece pay up.
Cost of our borrowing went up immediately.
How did that feel? Look, it was, first of all, it's not just that your cost of your borrowing is more.
It's also that you have to borrow much more because you realize that what you owe in a given year is bigger than what you thought you owed.
The real numbers showed that actually Greece owed a lot more than they thought, more than double.
And they were going to need to borrow to cover it.
And whenever you borrow a lot more, your cost of borrowing goes up.
And you know what else makes your cost of borrowing go up?
Having to come clean to your lenders about your numbers.
They don't like that. George found that trust had entirely evaporated.
Almost overnight. And once this happened...
Basically, the market said, OK, we don't trust your numbers.
And by extension, we don't trust any policy announcement.
So you're going to tell us that you're going to rein in the deficit.
We're not going to trust until we see it and then we verify it.
So you lose even the benefit of the doubt in everything that you're doing.
As with Argentina, people who no longer trusted the numbers just assumed the worst and charged Greece even more because of it.
And in Greece, they also had to fix the actual problem that the prior administration had been trying to paper over in the first place.
They had negative money and the market had lost trust in them.
So Greece had to go to the other members of the Eurozone, Germany, France, all those guys, and ask for money to fill in the hole.
Billions and billions of dollars. But as with all money, this was not free.
They had to agree to cut spending and raise taxes.
Austerity. Something no politician wants.
It generally means you've got a recession.
And for Greece, it did. For nearly 10 years.
The consequences of book cooking were really bad.
That loss of trust, which can happen very quickly and very easily, it is then incredibly difficult to bring back.
You lose it easily. And to gain it back, you've got to really prove yourself and go through hoops over years, have external agencies validate indeed your data.
And people don't realize how delicate this is.
Because messing with the numbers doesn't actually let you hide from reality.
Reality is still there. You're just delaying dealing with it and also making it worse.
Cooking the books ends up being more expensive than just dealing with the books in the first place.
So this past Friday, August 1st, when the news came out about the BLS head getting fired, has this been giving you flashbacks?
Oh, totally. To me, this is 2009. I mean, here is an economy which is running a very high deficit, debt and deficit, with an administration who likes to pretend that the normal laws of economics do not apply.
And suddenly it gets weak employment numbers.
So what does it do? It shoots the messenger.
And this is really bad because it destroys confidence into independent institutions.
And once you go that road, if we lose trust in these kind of numbers, if we lose trust in the data, if we lose trust in the people who produce those numbers, then there's nothing to base ourselves on when we have a political debate.
We can disagree on policies, but we should not be able to disagree on the underlying numbers.
The numbers should be the numbers. And then you can decide whether X policy is better than policy Y, right?
But the numbers should be the numbers that we all believe in.
So that there's some shared reality and then we can decide what to do about that reality.
And if we destroy that shared reality, then there is no basis for a rational conversation.
Then it all becomes points of view. What's going to happen next time the Bureau of Labor Statistics announces unemployment figures?
Will people say, no, unemployment is much higher?
No, it's much lower, depending on their political affiliation.
Once we lose that starting point and we accept that reality is a very relative concept that can be manipulated at will, then we are on the road to oblivion.
George says the U.S. is not yet on the road to oblivion.
But we may be at a crossroads. Right now, the data are, as far as we know, actually OK.
But maybe there's something to learn from Greece's experience.
Because the process of uncooking is unfun.
And if you can avoid it, you should. If you want to hear more, we and our siblings show The Indicator.
We have done a bunch of episodes about data and government statistics.
We did one about if you can just change how we measure GDP and others on how the BLS calculates inflation and the jobs numbers.
We put links to those episodes in our show notes.
Today's episode was produced by Sam Yellow Horse Kessler.
It was edited by Jess Jane. It was engineered by Jimmy Keeley with help from Kwesi Lee.
We got some fact-checking help from Sierra Juarez.
Alex Goldmark is our executive producer.
Special thanks to Hansi Lo Wang, Steve Pearson, and Nancy Potok.
I'm Mary Childs. And I'm Sally Helm. This is NPR.
Thanks for listening.
