Category: Economics

Malthus and Ricardo, Wages and Rent


Ferdinand Lassalle’s Iron Law of Wages, following from Malthus, and David Ricardo’s Law of Rent are some of the very first relatively quantitative attempts at statements or observations of economics and can IMHO be considered a sort of ancestor of modern economics. 

In the Iron Law of Wages, as population increases, the labor supply increases and thus the wage price decreases – which does mean that we assume labor demand is unaffected by population and thus labor demand is effectively exogenous.  Wages continue to decrease until they hit subsistence levels for laborers.  A further decrease in wages is unsustainable as laborers will literally be unable to sustain themselves, which may cause a decrease in population.  A decrease in population, i.e. a decrease in the labor supply, pushes wages back up to the long-term level, which is the minimum subsistence level.  If the wage price is above subsistence level, population will rise (the assumption is that any wage above the subsistence level contributes to population growth) until the wage decreases to the subsistence level.

Malthus’s Iron Law of Population is the observation that given enough food, population grows exponentially or geometrically while agricultural output – which is limited by 1. the amount of new land that can be put to agricultural use and 2. the amount of additional intensification that one can do to increase the output of existing agricultural lands, which Malthus understandably assumes to have diminishing returns – grows linearly or arithmetically.  For the former limit on agricultural output, his evidence is the population growth in the early United States where new land was plentiful (despite the existence of Natives on those lands) while his evidence for the latter limit on the diminishing returns of agricultural intensification is an appeal to common sense of the times (which may be understandable – we can suppose that it would be hard for someone in the early 1800s to think that agricultural output could grow to accommodate an exponentially growing population or that in the future, longer years of education would lead to declining fertility rates). Since linear growth has no hope of staying above exponential growth in the long run, Malthus’s conclusion is that once population hits the level where the masses can only afford a subsistence level of living, that will be the long run equilibrium for wages and quality of life. There may be ameliorating factors such as an increase in agricultural technology, delay in bearing children, and contraception, or direct decreases to population such as war and disease as such, but Malthus’s opinion was that none of that can overturn the Iron Law of Population. In any case, once population hits the level where people are living at subsistence levels, whether it be war, disease, or famine that keeps population from going above this long run equilibrium doesn’t change the fact that the factors that keep population from going above this equilibrium are painful to humanity.





The Terms of Trade of Brazil


An article in the New York Times by Paul Krugman talked about a current economic downturn in Brazil. What happened:

First, the global environment deteriorated sharply, with plunging prices for the commodity exports still crucial to the Brazilian economy. Second, domestic private spending also plunged, maybe because of an excessive buildup of debt. Third, policy, instead of fighting the slump, exacerbated it, with fiscal austerity and monetary tightening even as the economy was headed down.

What didn’t happen:

Maybe the first thing to say about Brazil’s crisis is what it wasn’t. Over the past few decades those who follow international macroeconomics have grown more or less accustomed to “sudden stop” crises in which investors abruptly turn on a country they’ve loved not wisely but too well. That was the story of the Mexican crisis of 1994-5, the Asian crises of 1997-9, and, in important ways, the crisis of southern Europe after 2009. It’s also what we seem to be seeing in Turkey and Argentina now.

We know how this story goes: the afflicted country sees its currency depreciate (or, in the case of the euro countries, its interest rates soar). Ordinarily currency depreciation boosts an economy, by making its products more competitive on world markets. But sudden-stop countries have large debts in foreign currency, so the currency depreciation savages balance sheets, causing a severe drop in domestic demand. And policymakers have few good options: raising interest rates to prop up the currency would just hit demand from another direction.

But while you might have assumed that Brazil was a similar case — its 9 percent decline in real G.D.P. per capita is comparable to that of sudden-stop crises of the past — it turns out that it isn’t. Brazil does not, it turns out, have a lot of debt in foreign currency, and currency effects on balance sheets don’t seem to be an important part of the story. What happened instead?

Slowly going over the three points that Krugman made in the beginning:

1. Commodity prices went down and Brazil exports a lot of commodities.

Brazil’s exports in 2016:


At a glance, we have among commodities: vegetable products, mineral products (5% crude petroleum, 10% iron and copper ore), foodstuffs, animal products, metals, and precious metals. Though picking out these may be over or underestimating the true percentage of commodity exports among all of Brazil’s exports, let’s use these for our approximation. The total percentage of these products is about 60%, where around 36% are agricultural commodities, around 27% are metal commodities (metals + iron and copper ore), around 5% is crude petroleum, and around 2% are precious metals. These categorizations that I did are improvisational and not following any definitions – they are simplifications.

Looking at the S&P GSCI Agricultural & LiveStock Index Spot (SPGSAL):


we definitely do see a downtrend in the last several years in agricultural commodities.

Looking at the S&P GSCI Industrial Metals Index Spot (GYX):


there was a decline from 2011 but a rise from 2016.

Looking at the S&P GSCI Precious Metals Index Spot (SPGSPM):


it’s been flat since around 2013.

Looking at S&P GSCI Crude Oil Index Spot (G39):


it has been low after a decline in 2014 with volatility in 2017-2018.

But instead of eyeballing this phenomenon with a bunch of different charts, there’s a way that can mathematically eyeball this in one chart, called the terms of trade.

Investopedia’s definition of terms of trade:

What are ‘Terms of Trade – TOT’?

Terms of trade represent the ratio between a country’s export prices and its import prices. The ratio is calculated by dividing the price of the exports by the price of the imports and multiplying the result by 100. When a country’s TOT is less than 100%, more capital is leaving the country than is entering the country. When the TOT is greater than 100%, the country is accumulating more capital from exports than it is spending on imports.

But how exactly do you calculate the “price of exports and imports” of a country like, say Brazil, that has USD 190B exports a year and surely thousands if not more different products, and what to do about the changing quantities of each of those products every year? How do we understand the terms of trade in a way that doesn’t vaguely seem like the current account balance? (which is the total value of exports minus imports, or net value of exports: \( EX – IM = \sum_{i}^{}{p_i \cdot q_i} – \sum_{i}^{}{p’_i \cdot q’_i} \) where \( p_i\), \( q_i \) is the price and quantity of export product \(i\) and \( p’_i\), \( q’_i \) is the price and quantity of import product \(i\).

The answer is by deciding on a base year to compare the year in question. For example, for the prices of products in the year in question, we sum the values of exports for each product in that year, i.e. \( \sum_{i} {p_{i,n} \cdot q_{i,n}} \) where \(i\) is the index for each different product and \(n\) is the year in question. For the prices of products in the base year \(0\), we take the price of each product \(i\) in that base year multiplied by the quantity of that product \(i\) in the year in question \(n\). In other words, we fix the quantity of each product \(q_i\) to the quantity of each product in the year in question \(q_{i,n}\) so that we are strictly comparing prices between year \(n\) and \(0\) and not letting changes in quantity \(q\) get in the way. This is the Paasche index.

Another way we can do this is: for the prices of products in the year in question \(n\), we sum the prices of each product in that year \( p_{i,n} \) multiplied by the quantity of each product from the base year \( q_{i,0} \), and for the prices in the base year \(0\), we take the price of each product \(i\) in that base year multiplied by the quantity of that product \(i\) also in the base year \(0\). So this time, instead of fixing the quantity of each product in the year in question \(n\), we fix the quantity of each product to the base year \(0\). This is the Laspeyre index.

Paasche index:

$$ P_{\textrm{Paasche}} = \frac{\sum_{i}{p_{i,n} \cdot q_{i,n}}}{\sum_{i}{p_{i,0} \cdot q_{i,n}}} $$

Laspeyre index:

$$ P_{\textrm{Laspeyre}} = \frac{\sum_{i}{p_{i,n} \cdot q_{i,0}}}{\sum_{i}{p_{i,0} \cdot q_{i,0}}} $$


Thus, by using such a price index calculation we “cancel out” the effect of changing export or import quantities so that we are only looking at the change of price of exports or imports between two time periods. With a base year \(0\), we can calculate the price index for exports in year \(n\), the price index for imports in year \(n\), and then divide the former by the latter to achieve the terms of trade for year \(n\):

$$ \textrm{Terms of Trade} \ = \frac{P_{\textrm{Paasche, exports}}}{P_{\textrm{Paasche, imports}}} \ \textrm{or} \ \frac{P_{\textrm{Laspeyre, exports}}}{P_{\textrm{Laspeyre, imports}}} $$


A terms of trade chart quantitatively summarizes all the above eyeballing we did with the visualization of Brazil’s exports and the charts of commodities indices as well as the eyeballing we didn’t do with Brazil’s imports. And we see what we expect in the above graph, which is a drop in Brazil’s terms of trade in the last several years.

What is the conceptual difference between the

\( \textrm{Terms of Trade} \ = \frac{P_{\textrm{Paasche, exports}}}{P_{\textrm{Paasche, imports}}} \ \textrm{and} \)

\( \textrm{Current Account} \ = \sum_{i}{p_{i,n,exp} \cdot q_{i,n,exp}} – \sum_{j}{p_{j,n,imp} \cdot q_{j,n,imp}} \ \textrm{?} \)

2. Brazil’s consumer spending declined due to rising household debt (the red graph):


3. Brazil implemented fiscal austerity to try to deal with “long-term solvency problems” and raised interest rates to try to deal with inflation, which was caused by depreciation in the currency. The currency depreciated due to lower commodity prices, which of course is also reflected in the terms of trade graph above.

Depreciating currency (blue) and inflation (change in or first derivative of red):


Interest rates raised to combat inflation:


We can see that interest rates rise in late 2015 as a response to rising inflation. Inflation drops as a response in the next couple of years, but this rise in interest rates contributed to the slow down in Brazil’s economy.


So we have a drop in the terms of trade (due to a drop in commodity prices), a drop in consumer spending (due to a rise in household debt in preceding years), and then fiscal austerity and monetary contraction as government policy responses, causing a recession in Brazil.

Portfolio Insurance and Black Monday, October 19, 1987

On the thirtieth anniversary of Black Monday, the stock market crash of October 19th and 20th in 1987, there have been mentions of “portfolio insurance” having possibly exacerbated the crash.


Portfolio insurance, in principle, is exactly what you might expect it to be: if you own a stock, Stock A, you insure it with a put option on Stock A.  Your position becomes equivalent to a call option on Stock A until the put option expires, with the price of this position being the premium of the put option when you bought it.

If you are managing a portfolio on behalf of clients, though, and you just need to insure the portfolio up to a certain date, after which, say, you hand over the portfolio, then to buy American put options to insure the portfolio would be unnecessary.  European put options would suffice.  So let’s suppose that we are only interested in European options.

In the article that I cite at the bottom (Abken, 1987), it seems that at the time, buying put options as insurance had a few issues.  This is assuming that the portfolio we want to insure is a stock index: the S&P 500 index.  The issues were:

  • Exchange-traded index options only had matures up to nine months
  • Exchange-traded index options had a limited number of strike prices
  • It’s implied that only American options were available (which we would expect have a premium over European options).

Thus, instead of using put options to insure the portfolio, the portfolio and put options are replicated by holding some of the money in the portfolio and some of it in bonds, Treasury bills, that we assume to provide us with the risk-free rate.

Without worrying about the math, the Black-Scholes equation gives us a way to represent our stock index S and put options P as:

$$S + P = S \cdot N_1 + K \cdot DF \cdot N_2$$






Abken, Peter A.  “An Introduction to Portfolio Insurance.”  Economic Review, November/December 1987: 2-25.

Link to articleArchived.


Value-added Tax and Sales Tax

(This is mostly a summary of and heavily borrowed from, archived).
(The first three figures are taken from Wikipedia).


Comparing No Tax, Sales Tax, and VAT

Imagine three companies in a value chain that produces and then sells a widget to a consumer. The raw materials producer sells raw materials to the manufacturer for $1.00, earning a gross margin (revenue – Cost Of Goods Sold, COGS) of $1.00. The manufacturer sells its product, the widget, to the retailer for $1.20, earning a gross margin of $0.20. The retailer sells the widget to a non-business consumer (for the customer to use and consume) for $1.50, earning a gross margin of $0.30.

No tax example

Imagine we add a sales tax of 10%.  Sales tax applies only to the transaction with the final end-user, the consumer, i.e. the final transaction. So the consumer pays the retailer $1.50 + 10% sales tax = $1.65 to the retailer. The retailer remits the sales tax, $0.15, to the government.

Sales tax example

  • Only retailers remit collected sales tax to the government. Different regions, products, and types of consumers may have different sales taxes, so retailers are burdened with the maintenance of functions that process this for every different kind of sales tax.
  • There are cases (e.g. in the U.S., remote sales, i.e. cross-state or internet sales) where the retailer isn’t required to charge sales tax on its sales to consumers.  Instead, the consumer is responsible for remitting a use tax to the government on his or her remote purchases.
  • Only end-users pay the sales tax. Thus, someone who is an end-user has an incentive to masquerade as a business and purchase products for usage.
  • The government thus requires businesses (namely non-retailers, in this example) with the burden to prove, via certifications, that it is a business (and thus does not need to pay sales tax on the products it buys) and that it sells to other businesses (and thus does not need to charge and remit sales tax on products it sells).

Now let’s take away the 10% sales tax and add a 10% VAT. The raw materials producer charges $1.00 + a $0.10 VAT, which is 10% of the $1.00 value they added to the product) to the manufacturer. It remits the $0.10 VAT to the government. The manufacturer sells its product to the retailer for $1.20 + 10% or $0.12: $0.10 of which is the VAT that the raw materials producer charged the manufacturer and is now getting “paid back” by this transaction and the remaining $0.02 of which is 10% of the value added to the product by the manufacturer, which is $1.20 – $1.00 = $0.20, and remits the $0.02 to the government. The retailer sells its product to the customer for $1.50 + 10% of $1.50 or $0.15 ($0.12 of which is VAT that it paid to the previous 2 companies in the value chain and is now being “paid back” by this transaction and the remaining $0.03 of which is 10% of the value that the retailer added to the product, $0.30) and remits $0.03 to the government.

VAT example

  • From the consumer’s viewpoint, nothing has changed. End-users still pay the same $1.65 for the widget.
  • The government earns the same $0.15 as it earned with sales tax.  But instead of receiving all of it from the final transaction between the retailer and the consumer, it earns it in bits of [10% * each value added by each company in the value chain], which are $0.03, $0.02, and $0.10.
  • From the perspective of each business, they’re charged VAT by companies that they purchase from and they charge VAT to companies/consumers that purchase from them.  When they’re charged VAT on purchases ($0.12 for the retailer in the example), they are effectively charged the VAT of all companies that are further down the value chain from them ($0.10 for the raw materials producer and $0.02 for the manufacturer).  When they charge VAT on their sales, they effectively charge VAT for the value they added ($0.03) plus the VAT of all companies that are further down them in the value chain ($0.12).  The difference, the VAT for the value they added, is remitted to the government ($0.15 – $0.12 = $0.03).  So when a company purchases and is charged VAT, they are effectively “in the red” for that amount of VAT ($0.12) until they can sell their product up the value chain and charge that amount of “downstream” VAT + the VAT they own on the value they added ($0.12 + $0.03 = $0.15).  Then with that sale, they get “refunded” the portion of VAT that they paid before ($0.12) and remit the remainder ($0.03) to the government.
  • All buyers, whether they’re a consumer or a business, pay the VAT. So there is no incentive for anyone to masquerade as anyone else (e.g. a consumer to masquerade as a business).
  • All businesses process VAT (charged VAT on products they buy, charge VAT on products they sell, and pay VAT to the government) so all businesses are burdened with the maintenance of functions that process this.
  • Because businesses are charged VAT when they buy products and remain “in the red” that amount of VAT until they can sell those products, they are incentivized to make sure they charge VAT on the products they sell in order to make up that VAT they were already charged (e.g. the manufacturer paid the raw materials producer $0.10 of VAT so it’s incentivized to charge VAT on the products it sells to the retailer to make sure to make up for that $0.10). Because everyone is incentivized to charge VAT on their buyers, “everyone collects the tax for the government.”
  • This symmetry where everyone in the VAT system charges and is charged VAT doesn’t exist when it comes to cross-border trade, which is discussed below.


Sales Tax versus VAT

  • Most countries (166 out of 193 countries) in the world use VAT. The US uses sales tax and is the only one to do so in the OECD.
  • Asymmetry creates perverse incentives: In sales tax, only retailers charge sales tax and remit sales tax to the government. Consumers want to masquerade as businesses, retailers are not especially incentivized to make sure that their buyers are charged sales tax, and if there are ways to get around sales tax (remote sales, which include cross-state sales and online sales; wholesaling to consumers ), retailers and consumers might want to do that. In the US, retailers don’t need to charge sales tax to consumers buying in a state in which the retailer doesn’t have a physical presence.  In this case, there is a use tax charged on the consumer to make up for this, but compliance of use tax is low.  (Source, archived.)  Estimates of sales tax lost due to remote sales in 2012 varied up to a high of USD $23 billion where total retail sales that year (excluding food sales because many states don’t charge sales tax) was around USD $350 billion a month, i.e. around USD $4.2 trillion that year.  (Archived, archived, archived.)
  • In VAT, all businesses process VAT the same way, so there are no such perverse incentives.
    • The big exception to this is cross-border trade, i.e. imports and exports.  Governments have a choice of whether to charge VAT on goods it exports and goods it imports and whether to charge differently for every country it trades with.  This creates a huge potential for asymmetries in the VAT system.
  • Imports and Exports:
    • Sales Tax countries charge sales tax on imported goods if and when they reach the end-user.  If the imported good is exported again, then it hasn’t reached an end-user, and thus is never sales taxed.
    • Both sales tax and VAT are consumption taxes – the purpose is to tax consumption.  This is why the sales tax doesn’t tax a good that is imported and then exported without being consumed in the country.  VAT accomplishes this as well, but also would ideally keep the cross-border trade situation simple and sensible when dealing with other VAT countries or sales tax countries.
    • In order to have as symmetric and fair a system of cross-border trade, VAT countries generally:
      • Do not charge VAT goods that are exported.  When a good is exported by an exporter, the government refunds the exporter the entire VAT that it paid on its cost of goods sold purchases;
      • Charge VAT on goods that were imported on that good’s first subsequent sale that occurs after importation for the full sale value (not just the value added by the importer, which is [sale price – cost of goods sold], but the full [sale price]). I.e. after the importer imports the good, when that importers sells that good, that sales transaction is VAT-taxed for the full sale price.  This is assuming that the imported good is being sold to another domestic company and not being immediately exported.
      • Note that if an imported good is exported, the government does not receive any VAT.  This is the same as in sales tax and it accomplishes what a consumption tax is supposed to do (which in the case of a good that is imported and then exported without being consumed in the country is to not tax the good).  Each company in the value chain plays its usual part in the VAT system, but the last one, the exporter, is refunded by the government all the VAT it paid on its purchases of cost of goods sold.

Cross-border VAT

  • The reason VAT countries don’t tax their goods upon export is because sales tax countries don’t tax their goods upon export, so this keeps that part of the trade symmetrical.  This also prevents any case of a good being double-taxed during a cross-border trade.
  • The reason VAT countries tax their import goods (subsequent to the import transaction) is because if that good is going to be consumed in the country, not VAT-taxing it would mean the good would be untaxed during and after its cross-border transaction and thus have an advantage over similar competing domestic goods at this and every following point on the value chain since domestic goods have been VAT-taxed up to this point and will be VAT-taxed on all following points on the value chain.
  • The reason why an imported good’s subsequent sales transaction is VAT-taxed its full sales price instead of just the value added by the importer is because:
    • If the good is only taxed by its value-added amount instead, this still is not enough to offset the disadvantage of domestic goods (which have been VAT-taxed for all value that has been added to the product up to that point, not just the value added by the last company to sell it).
    • The government of the country in which the good is consumed ought, in principle, to capture the entire consumption tax on the good.  Thus, when the good went across the border and the exporting country’s government refunds the VAT to its exporter, the good is effectively “untaxed” at this point (the exporting country’s government has refunded all previous VAT on it and the importing country’s government has yet to tax any of the value that has so far been added by producers of the exporting country).  By VAT-taxing it by its full sales price after importation, the government of the importing country captures the VAT that the exporting country’s government refunded to its exporter or “resets” the VAT to where it ought to be at this point in the value chain for itself.
  • Missing trader fraud/Carousel fraud: A type of fraud that exists when a good is imported into and then exported out of a VAT country without the good being consumed in the country.  Since cross-border trade is a point of asymmetry in the VAT system, it makes sense that this is where fraud occurs.
    • Company A imports a good legitimately, paying the exporter EUR 100 for the good.  This transaction is VAT-free.
    • Company A sells the good to Company B for EUR 110.  This transaction is VAT-taxed its full sales price (since it is the transaction subsequent to the good being imported and also is not being immediately exported), and Company A owes the government this VAT.  Note that the good in this transaction is “new” to the country and thus the government has not received any VAT from this good further down the value chain (that VAT has been collected by the exporter’s government and refunded back to the exporter).  If the VAT is 20%, that 20% is charged on the full EUR 110 sale price of the transaction (not the value added by Company A, which is EUR 10).  The total price of the transaction with VAT to EUR 132 and the government expects to be remitted a VAT of EUR 22 from Company A for this sales transaction.
    • Company B exports the good.  This transaction is VAT-free.  Furthermore, Company B has paid Company A a VAT of EUR 22, so it is entitled to a refund of EUR 22 from the government as the good is being exported and not consumed in the country.
    • Company A disappears or goes bankrupt without paying the VAT (of EUR 22) on the sale of goods by Company A to Company B.  This is key to the fraud because Company A is supposed to pay VAT for the full sales price of its sale of the good (20% * EUR 110 = EUR 22), not just VAT of the value added (20% * EUR 10 = EUR 2).  In the diagram above that depicts cross-border trade, the retailer would owe the government $0.15 of taxes, not $0.03 as in the diagrams that are further above that don’t depict cross-border trade.  By Company A disappearing, the government is losing the VAT of all value that has been added to the good by companies from this point and all the way down the value chain.
    • With no fraud occurring, the government is supposed to earn 0 tax: charge VAT starting from the importer selling the good to domestic companies but then refund all that VAT to the exporter at the end who exports the good since the good is not to be consumed in the country.  But instead, in this case where Company A disappears, the government has lost EUR 22 by refunding Company B, the exporter, for the VAT that it paid on its purchase of the good.
    • In reality, if the importer (Company A) and the exporter (Company B) are working together, the good may never even physically leave the port, and is imported, sold, and exported only on paper.  If there are many companies in between Company A and Company B (e.g. it could instead by A -> X -> … -> Z -> B), it could be difficult for the government to prove any wrongdoing by Company B as the link between A and B will be weak (Company B may even be innocent in some cases where the only fraud is Company A disappearing without paying its VAT) and thus the government will be obligated to refund the VAT to Company B that it paid on its purchases.
    • According to sources found in Wikipedia, it’s estimated that the UK annually lost around GBP 2 billion in 2002-2003 (archived) and between GBP 2 billion and GBP 8 billion annually for the years (archive) between 2004 and 2006 due to this kind of fraud.  Total UK retail sales (archived) for these years was around GBP 250 million to GBP 300 million.  For the EU, 2008 estimates were EUR 170 billion lost (archived) due to this type of fraud.  Total EU-27 retail turnover in 2010 was around EUR 2.3 trillion (archived).
    • In the above link to a BBC article from 2006, it says that in order to combat the losses from this fraud, the government is implementing a new system where:

      Under the new rules the last company to sell on goods like mobile phones – such as a retailer – will be responsible for paying the VAT.

      So it sounds like the portion of VAT that the government is “missing” from the imported good will be paid by the retailer instead of the importer – in other words, it’s a bit like the sales tax system.  In this system that’s described, if a good is imported and then exported, the amount that the government would refund the exporter will be smaller than in the previous system, making the fraud much less damaging.  And if the retailer disappears without paying the taxes it owes, that’s the same as a retailer disappearing in a sales tax system without paying taxes, or a raw materials producer in a VAT system disappearing without paying taxes.  (These cases are a much simpler sort of tax evasion and unlike the missing trader/carousel fraud where the importer disappearing and not paying the “missing” chunk of VAT that the government is owed is in combination with the exporter that is “refunded” that chunk of VAT from the government, even though the government never received that chunk of VAT from the missing trader.)

Instead of taxing the importer the “missing” VAT, tax the retailer the “missing” VAT

In Country A, the government refunds the manufacturer/exporter the VAT that it paid on its purchases.  In Country B, the importer is charged VAT only on its value added, and that VAT is remitted to the government.  The retailer is charged VAT on its value added (10% * $0.20 = $0.02) and on the “missing” VAT from the value added to the product prior to importation, which is $1.20 (the price that the importer paid the manufacturer/exporter), so that comes to 10% * $1.20 = $0.12.

If the good is exported, refund the exporter as usual

If the good is imported into Country B and then exported without reaching a consumer, the amount of VAT that the importer is charged is only on its value added (10% * $0.10 = $0.01).  Thus, the amount of VAT that the government refunds the exporter is only that amount ($0.01).  If the importer disappears, the government only loses $0.01, which is 10% of the value added by the importer, not 10% of the full sales price of the product at this point.  Furthermore, if the importer and exporter work together and the importer sells the good to the exporter at a much higher price (raising the value added and the potential VAT that the government is supposed to refund the exporter), the exporter still needs to legitimately export the product in order to qualify for the refund.  It’s theoretically possible for the exporter to operate at a loss to make this possible, but this might raise an additional red flag that the government would become suspicious of, raising the risk of doing the fraud.


Economic Impact of Sales Tax versus VAT

Back to the diagrams for no tax, sales tax, and VAT:

No tax

Sales tax


Although only the consumer actually pays the tax in the end, as prices are raised at other transaction points, there is some friction that will discourage those transactions by some amount.

One can also think of this as an overhead cost for the businesses involved.  In the no tax and sales tax situations, the manufacturer buys goods at $1.00 and sells them at $1.20.  In the VAT situation, the manufacturer buys goods at $1.10 and sells them at $1.32, which minus the VAT remitted to the government becomes $1.30.  In both cases, the manufacturer earns a profit of $0.20, but there is an overhead of $0.10 in the VAT case.  An extreme analogy is: if you are a company that makes a profit of $1 on each good you sell, would you rather buy goods for $2 and sell them for $3 to earn your $1 profit or buy goods for $1,002 and sell them for $1,003 to earn your $1 profit?  Surely the former is easier and has less friction.

Back to the economic interpretation: if the demand and supply curve of a transaction point in a no tax situation is this:


then by adding a tax to the transaction, the price is increased.  For convenience, we add a second supply curve that is “supply + tax.”  While the end result is the same if we left the supply curve alone and added a “demand – tax” curve instead, the supply + tax curve more conveniently takes the hypothetical price of a product sold (the supply curve) and then adds the hypothetical price + tax of a product sold (the supply + tax curve).  What one can also do instead of drawing a new curve is take the vertical distance of the final tax per product and “fit it in between” the two curves from the left side of the diagram, and the end result will be the same.


(The diagram says “Consumer Surplus” but since this may represent a business-to-business transaction, it’d be clearer to just say “Purchaser Surplus.”)  So a tax will cause less quantity to be transacted, a higher post-tax price, some government tax revenue, lower purchaser and producer surpluses, and some deadweight loss.  Unless the government tax revenue is spent in a way that can overcome that deadweight loss (e.g. spending on things that have positive externalities), we have an inefficient outcome.

So in the VAT system, businesses are contending with higher prices (which is like more overhead) and lost quantity transacted compared to the sales tax system.  This is another cost that the VAT system pays (in addition to the missing trader/carousel fraud) in order to have a “symmetric” system where almost everyone in the value chain pays and collects VAT.


The Theory of Interstellar Trade

The Theory of Interstellar Trade, by Paul Krugman (1978)



Assume we have two planets, Earth and Trantor, separated by a large distance, the traversal of which necessitates travel at velocities comparable to the speed of light.  Assume that Earth and Trantor are in the same inertial reference frame, i.e. they are not accelerating with respect to each other.

Assume that a spaceship traveling between the two planets travels at a constant v\ .

Let’s say that from the perspective of an observer on one of the planets, the time it takes for a spaceship to make the trip is n\ .

Then, the time it takes for a spaceship to make the trip from the perspective of someone on the spaceship is


(which is shorter than n\ ).  The factor on the right is a well-known result from relativity, derived mathematically from the Lorentz transformation, that gives us “time dilation” from traveling at relativistic speeds.  Krugman demonstrates the above relation by representing the voyage in Minkowski space-time in a figure using imaginary axes.



p_E\textup{, } p_T be the price of Earth goods and the price of Trantorian goods on Earth, respectively.

p_E^*\textup{, } p_T^* be the price of Earth goods and the price of Trantorian goods on Trantor, respectively.

r\textup{, }r^* be the interest rates on Earth and Trantor, respectively.


The first question Krugman asks is what is the correct interest rate on a planet with regard to interstellar trade, since depending on whether you’re on a planet or on a spaceship engaging in trade, the passage of time is different.  So putting ourselves on Trantor, we compare what happens to a Trantorian trader who can engage in investing in a Trantorian bond or engage in interstellar trade with Earth.

Interstellar trade with Earth would involve buying goods on Trantor, traveling to Earth, selling the Trantorian goods there and then buying Earth goods with those proceeds, and then traveling back to Trantor and selling the Earth goods there.  Let c\ be the cost of outfitting the spaceship.  The cost of buying goods on Trantor is q_T^* p_T^* where we define q_T^* as the quantity of Trantorian goods to be traded.  Thus, the initial expenditure of the Trantorian trader while still on Trantor is

c + q_T^* p_T^* .

The trader then travels to Earth and then sells its (not his or her) q_T^* goods at price p_T .  The trader now has money q_T^*p_T and will buy Earth goods with this money to bring back to Trantor.  Earth goods cost p_E so the quantity of Earth goods that can be bought is q_T^*p_T/p_E .  Let us call this quantity of Earth goods to be brought over to Trantor q_E^* .  So we have

q_E^* = \frac{q_T^*p_T}{p_E}

When the trader arrives back on Trantor, it will sell these Earth goods at price is p_E^* , resulting in a revenue of

\frac{q_T^* p_T p_E^*}{p_E}

But the Trantorian trader also had the choice of investing its money in a Trantorian bond before leaving for Earth.  The expenditure needed for the trading venture of c + q_T^* p_T^* invested into a Trantorian bond would:

after time 2n , which is the time it takes for a round trip from the point of view of a trader who decided to stay on Trantor for that duration of time instead of go to Earth and back, become

(c + q_T^* p_T^*)(1+r^*)^{2n}


after time 2\overline{n}=2n\sqrt{1-\frac{v^2}{c^2}}, which is the time it takes for a round trip from the point of view of a trader who decided to actually make the round trip and thus be on a spaceship for that duration of time (maybe he puts some money into a Trantorian bond and some money into the trading venture that he personally goes on), become

(c + q_T^* p_T^*)(1+r^*)^{2n\sqrt{1-\frac{v^2}{c^2}}}

So which perspective is right?  The perspective from Trantor or from the spaceship?  Krugman answers this by reminding us of the reasoning behind present value calculations, which is that of opportunity cost – any money that you choose not to possess today (by delay possessing it until the future for a larger amount) is money that you could have invested in a riskless bond today that would have also grown in value in the future.  And so the way I’ve framed it here sort of answers the question in advance, which is that the correct value of the bond is from investing it risklessly with some riskless bond issuer, like the Trantorian government.  The opportunity cost of spending money to outfit a trading venture to Earth is the lost opportunity of buying a Trantorian bond with that money and then receiving the proceeds by the time the trader would return from the venture, which according to the bond issuer, the Trantorian government, is 2n time.  Thus, we have Krugman’s First Fundamental Theorem of Interstellar Trade:

When trade takes place between two planets in a common inertial frame, the interest costs on goods in transit should be calculated using time measured by clocks in the common frame, and not in the frames of trading spacecraft.

Another way to think about it is that a bond earns interest because the bond issuer gets to possess cash now, invest it during some time, and afterwards will have more cash as a result.  The location that this is occurring is at the bond issuer’s location, which is Trantor.  The trader can always come back to Trantor and earn his bond interest and principal (assuming it’s a bond that automatically renews after maturing every time).  All this occurs on Trantor, and so the bond’s time progression is according to Trantor’s time progression.


Now, for simplification Krugman assumes that perfect competition reduces the profits of Earth-trading to 0.  In other words, we force the revenue earned from a trading venture to Earth that started with expenditure (c + q_T^* p_T^*) to equal the revenue earned from if we had invested that same expenditure into a Trantorian bond and waited the time that it takes to make a round trip to Earth and back:

\frac{q_T^* p_T p_E^*}{p_E} = (c + q_T^* p_T^*)(1+r^*)^{2n}

Also for simplification, Krugman assumes away c\ , the cost of outfitting a spaceship:

\frac{q_T^* p_T p_E^*}{p_E} = ( q_T^* p_T^*)(1+r^*)^{2n}

\frac{  p_E^*}{p_E} =  \frac{p_T^*}{p_T}(1+r^*)^{2n} \hfill \textup{(Relative goods prices)}

Krugman asks (or rather, asks what if someone asks) about a Trantorian trader who instead of going back to Trantor, decides to settle on Earth, thus making a one-way trading trip.  Krugman writes that the Trantorian trader has two options: either buy Trantorian goods, bring them to Earth, and sell them there, or buy Trantorian bonds, bring them to Earth, and sell them there.  While in a “realistic” situation, we would just need to assume that there is a liquid market for Trantorian bonds on Earth to make the second option possible, let us restrict our picture to just imagining one Trantorian in the whole universe who wants to migrate to Earth and then as many other traders as we want to imagine who all only intend to make round trips and live on their home planet and thus have no interest in the bonds of the other planet (thus, no Trantorian bond market on Earth).  In this case, Krugman assumes that we just have at least one Earthling who wants to migrate to Trantor and thus may be interested in buying a Trantorian bond before making his or her one-way trip.

So pursuing option 2, what price could this Trantorian bond fetch on Earth?  The Earthling has a choice of either buying and then bringing Earth goods to Trantor for sale or buying this Trantorian bond instead that can be redeemed when he or she arrives on Trantor.  The value of a one “T$ (Trantorian Dollar) bond on arrival at Trantor will be (1+r^*)^{2n} in Trantorian Dollars.  Thus, we want to know what value of Earth goods, when sold on Trantor, would net the same amount of Trantorian Dollars, for that is the Earth price of the Trantorian bond.

Suppose the Earthling buys q amount of goods on Earth to bring to Trantor.  The cost of that purchase is p_E q in Earth Dollars.  Once on Trantor, those q goods are sold for a total of p_E^* q Trantorian Dollars.   Let this equal the amount that would have been earned if a Trantorian bond was redeemed instead.  Thus, we have

p_E^* q = (1+r^*)^{2n}

q = \frac{1}{p_E^*}(1+r^*)^{2n}

The original Earth price for this transaction was

p_E q = \frac{p_E}{p_E^*}(1+r^*)^{2n}

Thus, \frac{p_E}{p_E^*}(1+r^*)^{2n} is the fair price in Earth Dollars for a Trantorian bond that can be redeemed in Trantor for (1+r^*)^{2n} .

If the Trantorian trader went with option 1, that means investing that one Trantorian Dollar in buying Trantorian goods to sell on Earth instead of buying a Trantorian bond and bringing it over.  One Trantorian Dollar buys \frac{1}{p_T^*} quantity of Trantorian goods, which on Earth will sell for a total of \frac{p_T}{p_T^*} Earth Dollars.  But from the “Relative goods prices” equation, we have that

\frac{p_T}{p_T^*} = \frac{p_E}{p_E^*}(1+r^*)^{2n}

Thus, one Trantorian Dollar invested in Trantorian goods and brought over to Earth can be sold for an amount of Earth Dollars that is the same as the fair price in Earth Dollars that a Trantorian bond will fetch on Earth.  So as long as there is at least one Earthling who also wants to make a one-way trip to Trantor and is open to buying a Trantorian bond instead of buying and bringing over Earth goods to sell, the Trantorian trader is indifferent between bringing over Trantorian goods to sell or bringing over Trantorian bonds to sell – both will earn the same profit.  This shows that the First Fundamental Theorem of Interstellar Trade holds as long as for one migrant going one way, we have another migrant going the other way.  What we seem to have is that as long as there is an “effective” round trip (either made by one Trantorian trader, or one Trantorian and one Earthling migrant) and the assumption of no arbitrage (so on each leg of the trip, the traveler is indifferent between carrying goods or bonds), we have a relation between the Trantorian interest rate and the prices of goods, and that none of this challenges the statement that the Trantorian bond’s interest gains run according to Trantorian time (the First Fundamental Theorem).  For Earth, we can construct the same scenarios except with the trip originating from Earth.  Krugman writes that he proves the theorem in the presence of transportation costs in a paper from the future.

Krugman then asks whether interest rates on Earth and Trantor will be the same or not.  The “Relative goods prices” equation (and its associated assumptions) are kept.  Even though transportation costs, as in c , the cost of outfitting a spaceship, was assumed away earlier, the “Relative goods prices” equation shows that there is an effective cost to transporting goods across stars that comes from the transportation time needed.  (If you are on Earth and you desire Trantorian goods, even if it costs no money for the Trantorian trader to outfit its spaceship to make the journey to Earth and sell its goods to you, the profit that the Trantorian trader makes by the time it returns to Trantor after its round trip needs to match the profit it would have made from just investing in Trantorian bonds and sitting at home for the same duration of time.  Thus, that gets built into the cost of Trantorian goods in planets that are interstellar distances away from Trantor.)  Could interstellar distances cause differences in the planets’ interest rates?

We imagine a scenario where a Trantorian trader buys Trantorian goods, travels to Earth, sells the goods there, invests the proceeds into Earth bonds, spends k time on Earth, redeems the bonds and buys Earth goods with that money, travels back to Trantor, and sells the goods there.  This scenario involves the prices of goods and Earth’s interest rate; and there is always another option: invest in Trantorian bonds at the beginning, sit at home for 2n+k time, and then redeem after that.  Then by forcing a no arbitrage condition, we are then able to obtain a relation between the prices of goods, Earth’s interest rate, and Trantor’s interest rate.

In the first option, the trader buys p_T^* q_T^* (where q_T^* is the quantity of Trantorian goods bought) worth of Trantorian goods, travels to Earth, sells these goods for p_T q_T^* , invests that in Earth bonds for k time, earns p_T q_T^* (1+r)^k $ after that, buys \frac{p_T q_T^* (1+r)^k}{ p_E} quantity of Earth goods, travels back to Trantor, and then sells these goods for a total value of \frac{p_E^* p_T q_T^*}{p_E}(1+r)^k Trantorian Dollars.  Forcing a no arbitrage condition, we require that this revenue equal the revenue from investing the initial expenditure in Trantorian bonds for the same amount of total time, giving us:

p_T^* q_T^* (1+r^*)^{2n+k} = \frac{p_E^* p_T q_T^*}{p_E}(1+r)^k

(1+r^*)^{2n+k} = \frac{p_E^*}{p_T^*}\frac{p_T}{p_E}(1+r)^k

From earlier, we had

\frac{p_E^*}{p_E} = \frac{p_T^*}{p_T}(1+r^*)^{2n} \hfill \textup{(Relative goods prices)}

Putting the two equations together gives us:

(1+r^*)^{2n+k} = (\frac{p_E}{p_T}(1+r^*)^{2n})\frac{p_T}{p_E}(1+r)^k = (1+r^*)^{2n}(1+r)^k

(1+r^*)^{k} = (1+r)^k

r = r^*

Thus, we have the Second Fundamental Theorem of Interstellar Trade:

If sentient beings may hold assets on two planets in the same inertial frame, competition will equalize the interest rates on the two planets.

In his conclusion, Krugman writes

I have not even touched on the fascinating possibilities of interstellar finance, where spot and forward exchange markets will have to be supplemented by conditional present markets.

I was thinking this was a light-hearted paper on a light-hearted topic but looks like there’s a lot more out there.  Thirty-five years later, Krugman gives a mention (archived) on the topic.