« What makes you happy? | Main | Natasha's question »
Fiscal policy and the fetishization of measured gdp
Fiscal policy can raise measured gdp without improving the economy or human welfare. Take a polar case where there is zero crowding out and there is one unemployed worker. Government taxes a rich man who is initially hoarding liquidity and spends that money putting the unemployed man to work.
Measured gdp goes up but is the new state of affairs better than the old?
Say the unemployed man had valued his leisure at 20K and was not taking the previously available jobs for 19K. The government now pays him 40K. For the new worker that is a gain of 20K in value but of course the cash itself is a transfer and does not represent a net gain from an economic point of view (though you may like the distributional effects).
Let's say the new public sector output is produced by labor only. 40K is spent to produce output which, in a market setting, would be worth $20K.
Put aside money flows. We have 20K more of output value and a 20K sacrifice of leisure value.
That's no net gain, but measured gdp has risen by 40K. Note that the contribution of the public sector to measured gdp doesn't very effectively measure true value added.
Introducing some crowding out makes fiscal policy look worse, as would considered the deadweight loss from taxing the rich man. You could make many other adjustments to the example; some would favor fiscal policy, others would not. The most obvious adjustment to make fiscal policy look good is to assume that the new output a valuable public good. Or you might argue that the resulting spending multiplier is higher than the real balance effect which results from the rich man holding the money (lower prices for everyone else).
But keep this example in mind the next time you see a measured empirical connection between fiscal policy and gdp. "Money spent" and "gdp registered" are not the same as "value created."
Posted by Tyler Cowen on December 5, 2008 at 07:04 AM in Economics | Permalink
Comments
Repost...
The effect also works the other way round: when people cite numbers how lower taxation encourages people to work and leads to higher GDP, the numbers ignore that people are trading free time for income, with the net profit in welfare much lower than the increase in GDP.
I always thought this explained to some extend why the income gap between the US and the richer parts of Europe doesn't feel as large as it looks on paper
Posted by: Zamfir at Dec 5, 2008 7:31:50 AM
Fiscal policy can raise measured gdp without improving the economy or human welfare.
Indeed so. GDP is a measure of economic activity, not of wealth or any other utility function.
Zamfir: I always thought this explained to some extend why the income gap between the US and the richer parts of Europe doesn't feel as large as it looks on paper
Another way of putting it is that Europeans tend to prefer to spend more of their wealth buying leisure time instead of objects.
Posted by: Cabalamat at Dec 5, 2008 7:54:38 AM
" 20K and was not taking the previously available jobs for 19K. "
My god... do you truly believe that's what unemployment actually is? Like these people enjoying their time at home watching TV and spending time with their kids.
what about the effects of unemployment on all sorts of things that are vaguely related to hapiness, such as mental state (unemploymeny => depression), loss of health insurance, divorces, financial distress. I mean, if you want to argue that leisure time is not included in GDP calculations you cannot neglect these either.
And who said these 19K jobs are available in the first place?
Moreover, when you say
" 40K is spent to produce output which, in a market setting, would be worth $20K. "
So what? who said that the market is right?
Do you truly believe that there is no market failure going on in the economy, and that and kind of intervention that stimulates the economy is artificial and unproductive?
Posted by: Rafael at Dec 5, 2008 8:20:08 AM
"the numbers ignore that people are trading free time for income"
Leisure time has no monetary value, it only yields utility. The effect of lower taxation on the willingness of those previously idle to take up more employment does increase GDP and add value outside of a person's own economic situation, provided they are employing their time at work efficiently.
Posted by: Tom M at Dec 5, 2008 8:30:36 AM
You don't mention the lost utility of the wealthy taxpayer. If you're doing welfare economics you have to aggregate utility of all the population.
So +20K in output -20K in lost leisure is a wash but the poor agent has an extra 20K so if the disutility to the rich agent is less then 20K the economy is better off. And this is an economic (though distributional) argument.
Posted by: Jim in Ottawa at Dec 5, 2008 8:40:56 AM
This is going to sound snotty, but I mean it seriously, is this a novel concept in the economic community? Or is it the economists that smirk when politicians and bureaucrats toss around GDP numbers?
I've long thought activity is not synonymous with prosperity.
Posted by: Andrew at Dec 5, 2008 8:49:24 AM
Another way of putting it is that Europeans tend to prefer to spend more of their wealth buying leisure time instead of objects.
The problem with this claim (which is often made) is that it seems to assume (a) there is a one-to-one mapping between individual preferences and government policy, and (b) that all individual preferences over all other individuals in the society are identical. Given different public policies, the fact that Europeans work less than Americans doesn't tell you anything about the preferences of individual Europeans vs. individual Americans.
If you're doing welfare economics you have to aggregate utility of all the population.
A better theoretical criticism of welfare economics I have not heard. Utility functions are neither a primitive nor unique (only unique up to a monotonic transformation, which is fine if you want to talk about individual welfare but definitely NOT fine if you're trying to add them up. Yes, you need to do this to do public finance economics, but it's still not very meaningful.
Posted by: ryan yin at Dec 5, 2008 9:18:09 AM
The reason you want economic activity is not to keep the idle off the streets, but to increase the standard of living.
And it's all based on disagreements over value. My labor is worth less to me than what you offer to me in exchange for it, and vice versa. In every voluntary transaction, both sides disagree about values, and the difference goes into profit for both sides. That's why the transaction happened. Division of labor makes differences about values sharper yet, and that's why the division of labor helps.
The more voluntary trading that goes on, as a result, the higher the standard of living of the nation.
There's no such necessary effect for transactions that are not voluntary.
Posted by: Ron Hardin at Dec 5, 2008 9:23:30 AM
Ryan yin, people, especially managers, have to make decisions every day based on vague impressions of 'benefit functions' that are not quantified and might be impossible to quantify.
After all, not doing anything is a decision too. In case of welfare functions: if you base economic and fiscal policy purely on GDP or other measurable numbers, you have still made an implicit choice about welfare functions, and not one that is obviously better than the 'meaningless' other alternatives.
Posted by: Zamfir at Dec 5, 2008 9:34:36 AM
This comment relates partly to the present post and partly to previous discussions on World War II and recovery from the Great Depression. I've always been troubled by glib assertions by Keynesians that WW II brought recovery from the GD and "proves" the efficacy of fiscal policy. Military spending at WW II levels crowds out production of consumer goods, resulting in forced savings on the part of the public. People accept that in wartime out of patriotism, but it doesn't provide a model for peacetime policy. On the other hand, the surge in military spending after the Fall of France and before the U.S. entered the war did evidently have a stimulating effect. Defense spending went from 1.6% of GDP in 1939 to 2.5% in 1940 to 11.3% in 1941. Real personal consumption expenditures which had increased by 5.6% in 1939 and 5.2% in 1940 increased by 7.1% in 1941, but declined by 2.3% in 1942 (though it did increase again in the next two years at a modest pace). In short, military spending before Pearl Harbor did have a stimulatory effect without crowding out, but not after. I do think though that saving the world from Fascism was a "valuable public good".
Posted by: Phil P at Dec 5, 2008 9:55:10 AM
Until recently, I would have considered Tyler_Cowen's post obvious and uninteresting. My position used to be that, yes, GDP isn't a perfect measure, but can capture relative economic performance to a decent approximation, so long as you understand its limitations.
However, now I see how harmful it is to focus on GDP, and why it should be more of a footnote when talking about economies.
It ultimately measures how many times money changes hands, regardless of whether that purchase was sensible, whether it resulted from a new inefficiency arising, whether it's merely rebuilding damage, etc. So then when economists from the government and their sympathizers consider ways to "improve the economy", they're always thinking in terms of "how can me make this artificial measure go up?" which leads them to do things that have little to do with the *genuine* economic improvements Tyler_Cowen mentioned.
That's why we're in the bizarre position of the government pondering how it can make people act like idiots, seriously believing that "the economy" depends on that, and not bothering to render any judgment about an economy so predicated.
Economists of the world: if people want to produce *and* consume less, that's okay! Really!
Posted by: Silas Barta at Dec 5, 2008 10:05:48 AM
I think this was part of Keynes's point in talking about national dividend.
Raphael,
I know, I know. Almost compeletely tone deaf...
Posted by: mickslam at Dec 5, 2008 10:12:49 AM
Zamfir, it is certainly true that there is a lot of uncertainty in the world and we have to do the best we can with what we have. But the fact that we cannot prove a new policy is bad for welfare (because we can't define welfare) is no argument for the policy. I don't see why we have to be agnostic about state interventions in the economy; the burden of proof obviously has to be on advocates of intervention. Moreover, it's entirely possible to get a lot of mileage on just Pareto and near-Pareto improvements without specifying any particular welfare function. I recently read a optimal gas tax paper that gave rather specific numbers without assuming any sort of grand social welfare.
(As an aside, my criticism is not just epistemological or quantitative. It's not just that social preferences are poorly measured -- it's that they might not really exist. People have preferences. It's not clear why societies would. Call me a misanthrope, but just once, I would like someone to explain this clearly, or do something other than hand-wave at Arrow's Impossibility Theorem and say that we should ignore it because some nice policy would make ever so much more sense if there really was a General Will.)
Posted by: ryan yin at Dec 5, 2008 10:18:18 AM
Yours is essentially a model of voluntary unemployment (jobs available around the reservation wage), which hardly seems adapted to the current situation. The fiscal inducement to work just distorts behaviour, so no surprise it is welfare reducing. Picture something more realistic. The value of unemployed leisure is close to 0K, and we pay the guy to work and produce 20K. The net gain is then almost 20K (ignoring the loss to the money hoarder).
Posted by: Mark Bryan at Dec 5, 2008 10:36:54 AM
--"Money spent" and "gdp registered" are not the same as "value created."
So (forgive my ignorance) why isn't there a GVC (gross value created) index? Too difficult to measure? Too subjective? (How do those VAT countries determine value added?)
How about a continuing series, a la "Markets In Everything": "Examples of Pointlessness of GDP"
Posted by: Delirious at Dec 5, 2008 10:43:23 AM
This is a major flaw of GDP, the deflator doesn't come close to real inflation.
Posted by: aaron at Dec 5, 2008 10:46:06 AM
Tyler, you assume that the market value of the public good is $20K. I'm not sure why you chose this number. I suppose that if the market price for this public good were greater than $20K then the fellow would not be unemployed in the first place. That seems reasonable. However, if the worker produces a public good, how is it that the market can price the value of the good correctly in the first place? For sure, unless the externality were somehow corrected/internalized, the market price does not reflect the true value to society, which is definitely higher.
Posted by: samson at Dec 5, 2008 11:07:48 AM
As Matt and commenters here note, GDP has always been understood as not measuring aggregate welfare.
So my question is: is Tyler suggesting that the GDP/welfare mismatch is worse for the marginal government-spent dollar at current levels? Or do we have little evidence for that -- is the mismatch merely just as bad as the GDP/welfare mismatch for consumption, investment, and government spending at other margins?
Posted by: mk at Dec 5, 2008 11:08:01 AM
Fetishization of GDP also rationalizes Bastait's broken window fallacy. Car accidents and hurricanes force people to empty out their savings and put some more people to work, making GDP look better, but clearly welfare has not improved.
Posted by: mobile at Dec 5, 2008 11:09:53 AM
mk,
Then Yglesias' commenters are just flat out wrong if they are mean that it is generally understood. I think some people understand that (Cowen, for example), but the evidence suggests that most don't understand this. I have lost count of how many times I have encountered the "the destruction caused by Hurricane X will be good for economic growth going forward" argument propounded by journalist, politicians, and economists!.
Posted by: Yancey Ward at Dec 5, 2008 11:20:23 AM
Tyler provides a useful framework for thinking about whether we should increase public spending now.
Here's a slightly more interesting, and perhaps empirically-relevant version:
Suppose that the value of this particular public sector project is $30K (instead of $20K) but the other numbers are the same. Then overall welfare increases by $10K. What's interesting is that during a period of full employment, employing the worker at $40K would have crowded out $40K of employment in some other, private sector activity and presumably eliminated $40K of value. In that case, the public works would have been a BAD idea and destroyed value. However, when resources are sitting idle, funding the exact same project becomes a GOOD idea, creating value. This is true even if there is a zero multiplier. It's just a simple cost-benefit analysis with different opportunity costs in the two scenarios.
Bottom line: When there are slack resources, then many public sector project become welfare enhancing even if the identical project would have been wasteful during periods of full employment. By continuity, the more slack the resources, the more such project become beneficial. Gov't spending is more cost effective in recessions than booms because the opportunity cost is lower, and so it should increase.
Policy implication: Policy-makers should go down the list of projects they have been considering and fund the next projects on their cost/benefit list.
Posted by: a student of economics at Dec 5, 2008 11:41:45 AM
1) This is an unusually interesting and timely discussion, but with one big drawback --- which, it seems, influences too much economic theory when it comes to policy-guides for dealing with real-world problems like the current economic recession and financial credit-crunch . . . the first one of a system-wide sort since the early 1930s, and on a global level now as then.
The drawback? The real balance effects analyzed here are strictly deductive . . . beginning with Tyler’s two-agent model, influenced by the theoretical premises of Keynesian or Austrian or Chicago-school economics.
……
2) But wait.
Fortunately, we have a real-world example of whether the real-balance effects did replicate the circumstances that Pigou and others hypothesized theoretically in the 1940s when they developed their theoretical work about such effects. To wit: Japan’s experience with systematic deflation for over a decade between the mid-1990s and mid-2000+ decade . . . the only relevant example since the 1930s for an advanced industrial country. And it’s made all the more relevant and informative because first a zero-interest rate policy was introduced --- which seemed, on the surface, to coincide with a liquidity trap --- in 1999, followed by quantitative easing QE a year or so later.
And even more fortunately --- instead of analyzing data about what was really going on with Japanese consumption trends in the aggregate, derived from GDP trends --- there were systematic and recurring opinion-surveys of the expectations of Japanese consumers regarding the declining price level they faced and how it affected their willingness to purchase products at present or delay them into the future.
…….
3) The results?
According to the best systematic study of consumer expectations --- remember, actual ones (not hypotheticals) --- deflationary expectations actually discouraged household consumption . . . and especially durables, “by delaying the timing of purchases.” Their policy-recommendation after a lengthy analysis of the opinion data? “Deflationary expectations should be upwardly revised to restore a vital Japanese economy.”
Here is the conclusion of the study, which is worth quoting because of its relevance too to the effects of QE on Japanese price-expectations --- a discouraging impact.
Click Price Expectations and Consumption under Deflation: Evidence from Japanese Household Survey Data,
Masahiro Hori and Satoshi Shimizutani , June 2005
“Our estimates indicate that average price expectations ranged from minus 0.2 to zero
percent in 2001 and 2002, increased to 1 percent in the first quarter of 2003, declined to 0.2 percent
in the second quarter, and showed a steady increase toward 0.8 percent by the first quarter of 2004.
Price expectations depend on current price movements and lagged expectations.
“A series of quantitative easing monetary policies were not very effective in changing the price expectations,since the policy announcements caused revision of price expectations only for a small proportion,5-10%, of people surveyed. The increase observed in the first quarter of 2003 was a reaction to the outbreak of the Iraq war. Our study also confirms that deflationary expectations discourage household consumption, mainly durable consumption, by delaying the timing of purchase.
“The findings of this paper thus suggest that deflationary expectations have to be revised upward to stimulate household consumption. However, quantitative easing measures were not very effective in changing price expectations. Accordingly, policy authorities need to implement
policies to change expectations in a more aggressive and understandable way for general public for there to be any significant impact on household behavior.”
………
Michael Gordon, AKA, the buggy professor
P.S. For what it's worth, the major value of Keynesian economics in something like the current intensifying recession amid a global financial meltdown lies in its stress (forgotten in the IS-LM versions and Samuelson's and recent New Keynesian interpretations) on the huge uncertainties in the minds of economic agents . . . intensified by inter-temporal problems: consumers, wage-earners, savers, businesses and households seeking loans, and bond and equity-holders and would-be buyers or sellers.
Posted by: the buggy professor at Dec 5, 2008 11:44:16 AM
Unlike Tyler's analysis: under current conditions you are not transferring the marginal dollar from a rich, liquidity-hoarding guy. You are transferring the marginal dollar from someone 10 or 15 years in the future who is better off than we are.
So this is a case of "consumption smoothing." And if analysts are right about the psychological or coordination problem that panic begets panic, then this is not even "consumption smoothing" but more like taking some bricks from a section of wall that is not under attack, and redistributing them to a section of wall that is in danger of collapse, with the understanding that if the wall is breached anywhere, everyone on the inside is in danger.
Posted by: mk at Dec 5, 2008 11:55:52 AM
This argument depends on the government paying more than market rates and/or not getting its moneys' worth in increased productivity. But the same is true in the private sector; paying an executive more than they're worth is *also* a net loss to someone.
The solution is the same as in the private sector: hire efficiently and put people to work on important projects, not make-work. Hiring efficiently also employs more people with the same amount of money.
This is harder to judge without market discipline, but not impossible, and as the previous bubble showed, the private sector is no great shakes at it either.
Posted by: Brian Slesinsky at Dec 5, 2008 11:56:48 AM
Tangentially riffing off Brian's post:
What about giving everyone a "USA investment account", where you stick $10K in an account for every household, and they have to invest it in something (stocks or bonds). Use it or lose it: you don't invest, the money vanishes.
Is that likely to have a worse result than having the government spend $1T lending or giving money to companies (or public projects), according to the government's discretion?
In other words, is the government's role to solve a coordination problem (which might also be solvable by weird game-theoretical helicopter money drops) or is the government also endowed with clearer insight about the relative success of various potential projects?
Should we be thinking harder about game-theoretical helicopter drops of money?
Posted by: mk at Dec 5, 2008 12:26:18 PM
