Tuesday, February 28, 2006
Positional grammar (Use Unicode)
What a day. GDP was revised up to 1.6% form 1.1% from the 4th quarter, but the good news seemed to have ended there. Consumer confidence came in under expectations as well as the Chicago Manufacturing Survey report. The GDP chain-deflator grew 3.3% in the 4th quarter while the market only expected 3.0%. Inflation rears its ugly head sending indices downward. The Dow fell through the 11K mark for the first time since February 13th.
Classical Chinese is an odd animal. It's certainly possible to read it with modern Mandarin pronunciations, but for someone trying to decipher and translate the passages, it seems like knowing actual Mandarin is a handicap. At least that's what I've come to believe this past semester. The majority of my class consists of white graduate students who are much more adept than I in this endeavor of translation. We read the texts in Mandarin, Japanese, and Korean. The professor who guides us, the one a few of us playfully dub Mair-zi, would probably ask us to do it in Canto or Min too if one among us actually knew how to speak it.
The fact that classical Chinese sentences use implied subjects so much of the time makes it an interesting read. Beyond that, for the words that make up the predicate, parts of speech is determined by position. A word can be adjective, noun, verb, adverb depending on where it is in a sentence. Take a look a 師 (shi1) for example.
師言 - shi1 yan2 - adj - the words of the master/teacher/rabbi
道師 - dao4 shi1 - n - Taoist master
師之 - shi1 zhi1 - v - to make it one's master
Or take 足 (zhu2) for example. As a noun it means a foot or to be satisfied, which is also the default interpretation that a Mandarin speaker would give it. In classical however, it's also possible to 'foot' something/someone: 为之足 (wei4 zhi1 zhu2) For it, give feet to.
Using classical's rules for sentence formation can create some really bizarre abominations. Take this construed 7 character sentence:
人人人人人人人
It's a perfect classical sentence, but there no way a modern speaker of Mandarin can make sense of it. Maybe annotating the characters with pinyin may help.
ren2 ren2 ren2 ren2 ren2 ren2 ren2
Never mind. In hindsight, having typed it out didn't help much. Grouping the pinyin would definitely help break down the sentence into logical, lexical units.
renren ren ren renren ren
However, we are still short on clues. So mentally, stick in the punctuation.
renren ren ren, renren ren.
A direct translation would go something like this: [If] people people people, people people.
In spoken English, it would then be: [If] people treat as people all people, people then become people-like.
Who said learning Chinese was hard? Quite primitive if you ask me.
Classical Chinese is an odd animal. It's certainly possible to read it with modern Mandarin pronunciations, but for someone trying to decipher and translate the passages, it seems like knowing actual Mandarin is a handicap. At least that's what I've come to believe this past semester. The majority of my class consists of white graduate students who are much more adept than I in this endeavor of translation. We read the texts in Mandarin, Japanese, and Korean. The professor who guides us, the one a few of us playfully dub Mair-zi, would probably ask us to do it in Canto or Min too if one among us actually knew how to speak it.
The fact that classical Chinese sentences use implied subjects so much of the time makes it an interesting read. Beyond that, for the words that make up the predicate, parts of speech is determined by position. A word can be adjective, noun, verb, adverb depending on where it is in a sentence. Take a look a 師 (shi1) for example.
師言 - shi1 yan2 - adj - the words of the master/teacher/rabbi
道師 - dao4 shi1 - n - Taoist master
師之 - shi1 zhi1 - v - to make it one's master
Or take 足 (zhu2) for example. As a noun it means a foot or to be satisfied, which is also the default interpretation that a Mandarin speaker would give it. In classical however, it's also possible to 'foot' something/someone: 为之足 (wei4 zhi1 zhu2) For it, give feet to.
Using classical's rules for sentence formation can create some really bizarre abominations. Take this construed 7 character sentence:
人人人人人人人
It's a perfect classical sentence, but there no way a modern speaker of Mandarin can make sense of it. Maybe annotating the characters with pinyin may help.
ren2 ren2 ren2 ren2 ren2 ren2 ren2
Never mind. In hindsight, having typed it out didn't help much. Grouping the pinyin would definitely help break down the sentence into logical, lexical units.
renren ren ren renren ren
However, we are still short on clues. So mentally, stick in the punctuation.
renren ren ren, renren ren.
A direct translation would go something like this: [If] people people people, people people.
In spoken English, it would then be: [If] people treat as people all people, people then become people-like.
Who said learning Chinese was hard? Quite primitive if you ask me.
Monday, February 27, 2006
Houses, houses
New home sales figures came in today at 8:30 below market expectations. 1233K versus the market expectation of 1270K units. This is just another indication of the cooling housing market Bernanke as well as many other economists have alluded to. New home sales measure the number of new privately owned one-family houses sold. The number is very volatile, but it is definitely a pro-cyclical indicator as sales usually rise at the beginning of business cycles and decline as consumer's appetite for housing is sated.
Friday, February 24, 2006
Durables, durables
Durable goods orders came in at 8:30 this morning. It recorded at 10.2% decline month over month from December. Most of it could be traced to the glut of orders Boeing received for its Dreamliner and other aircraft in December. Those were a one-time orders that greatly inflated the numbers for December. The report has implications for long term production capacity despite the fact that it is very volatile.
Thursday, February 23, 2006
On CPI and ports (oh, the complex world of economics)
This feels like something only a MGMT concentrator would do. Please humor me, will you? After you are done with that one, follow up.
I've been behind on my daily posting due to the onset of midterms, heavy coursework, and my 3 jobs. In a burst of productivity, I finished the three missing posts including this one. Have fun!
The CPI rose by 0.7% month over month (mom), and is up 4% over the past 12 months, outstripping earnings growth of 3.6%. The obvious culprit was energy prices, which have risen 25% over the past year. The big question is how much and when those energy prices feed through into other prices. As the PPI released from last week indicated, producers have not been completely passing on the higher costs due to energy onto their customers. It takes time for the market to feel the effect of the part that is passed on. In the short run, this is due to to stickiness of prices, but once price adjusts we will see a change in the core CPI. The core rose only 0.2% mom, up a total of 2.1% over the past 12 months. Given last week's .4% increase in the core PPI, it seems to suggest that prices have not yet adjusted for consumers. This is good news since inflationary pressures will remain low, which means that stocks and bonds receive less discounting. Less discounting yields higher prices.
On another note, there has been a controversy brewing for awhile now concerning Dubai Ports World assuming management over some U.S. ports. Aside from the security issues inherent in the foreign ownership of real U.S. assets, especially in the post-9/11 world, there is the interesting issue of the U.S. current account deficit (CA) to consider.
Remember that the flip side of the CA deficit is the NX. As the U.S. imports more goods, dollars flow outward into the hands of foreign firms which then use those dollars to purchase U.S. assets. In the recent past, those assets were financial (bonds, treasury notes, etc.), but that money can also be used to buy real assets like ports. This is called foreign direct investment (FDI). In total, in 2004, foreigners held $1.9 trillion in U.S. stocks, $2.2 trillion in U.S. government bonds, $2.1 trillion in U.S. corporate bonds, and $2.9 trillion in U.S. bank loans, but only $2.7 trillion in U.S. real assets.
Emerging economies in particular have bought mostly financial assets in the US, especially bonds. China has been particularly fond of the 10 year U.S. treasury, so much so that part of the flattened yield curve can be attributed to their voracious appetite. Oil producing countries have also used their oil-windfalls to buy U.S. financial assets. In 2004, Middle-East oil exporters held a total of $121 billion in U.S. securities. Financially, countries, especially emerging markets, prefer receiving FDI in real assets, because that's harder to reverse. Investors do not have the same liquidity with FDI compared to financial assets. In the case of an economic downturn or political destabilization, it would be harder to pull out of a country when holding real assets like factories.
So is this a good thing? Aside from the obvious concerns over security, there is also the greater question of how long can the U.S. afford to have such a large CA deficit. At around 6% of GDP, this is not sustainable in the long run. When foreign asset holders decide to stop lending money to fuel U.S. purchases (and they eventually will), the U.S. will be forced to cut back and maybe even learn to save.
Finally, I think I'm beginning to understand it all.
Wednesday, February 22, 2006
Menu preferences, beginning thoughts
Consider this example: The host of a garden party hands a fruit tray to a guest, who can choose between a mango or two apples. The guest in general prefers mangos to apples, but does not want to pick the last mango. Hence, he chooses an apple. Yet he might be very pleased if someone else were to 'force' that last mango on him. Similarly, if there were two mangos in the tray, he might happily choose one of the mangoes.
From this example, we can see that there may be preferences which are dependent on the menu of choices available to the chooser. Given a slightly different menu selection, people may select outcomes that do not align completely with their desired preferences. I haven't yet thought of the interesting implications, but I'll ponder this one for a while and maybe write a longer post on it some day.
From this example, we can see that there may be preferences which are dependent on the menu of choices available to the chooser. Given a slightly different menu selection, people may select outcomes that do not align completely with their desired preferences. I haven't yet thought of the interesting implications, but I'll ponder this one for a while and maybe write a longer post on it some day.
Tuesday, February 21, 2006
Killing 'E' Part 2 (or the right way to consider covariates)
The Leading Indicators report was put out by the Conference Board earlier today at 8:30am. The report generally is not a major market mover for obvious reasons. Most of the values included in the calculations have already been released in the weeks prior. Analysts can generally predict with high degrees of accuracy what the numbers ought to be.
Despite my professed hatred for data mining and regression analysis, I will yield a bit. There may yet be a way to adequately incorporate variables into probabilistic models. In a very basic sense, probabilistic modeling is line fitting. It's just in our religion, we hold that our lines are superior, and they are.
Once again remember that probabilistic modeling begins with modeling the observable choice that agents must make. We believe that the choice follows a random process that can be described by some stochastic process dictated by some sort of rate parameter. This is the basic model, but obviously the world is far more complicated than that. The most blatant shortcoming of this simple model is that it leads us to assume that all the agents act according to a homogeneous rate. In the real world, there is inter-agent heterogeneity, that is, they perform the act randomly at varying rates.
In that case, let us define the rate parameter as a function of another stochastic process to represent how we believe the rate is supposed to be distributed over the population of agents. Putting this together with the simple model, we now allow for our agents to be heterogeneous with regard to the rates at which they perform such an action. This is one of the fundamental aspects of why probabilistic modeling can provide amazingly accurate predictions when properly applied. Given that we used another stochastic process to bring in heterogeneity into the rate parameter, can you think of an analogous way of doing something similar with covariates?
Consider this: instead of telling the story that the rate parameter is spread over a population governed by a stochastic process, why don't we say that the rate parameter begins at a default rate and moves up or down in accordance to certain variables values of each particular agent? Use regression analysis to conjure up the relevant betas and then plug those into a function to modify the rate parameter. The great thing about using this method is that it does not necessarily prevent you from bringing in a stochastic process for the rate parameter itself. It is possible to implement both systems in tandem. Of course, with more complexity, you sacrifice more degrees of freedom, but in most cases, "over-fitting" will not be a problem if the number of covariates is kept under control.
I'm going to play around with this and probably put something up in the WIPs section when the initial write-up and tests are done.
Despite my professed hatred for data mining and regression analysis, I will yield a bit. There may yet be a way to adequately incorporate variables into probabilistic models. In a very basic sense, probabilistic modeling is line fitting. It's just in our religion, we hold that our lines are superior, and they are.
Once again remember that probabilistic modeling begins with modeling the observable choice that agents must make. We believe that the choice follows a random process that can be described by some stochastic process dictated by some sort of rate parameter. This is the basic model, but obviously the world is far more complicated than that. The most blatant shortcoming of this simple model is that it leads us to assume that all the agents act according to a homogeneous rate. In the real world, there is inter-agent heterogeneity, that is, they perform the act randomly at varying rates.
In that case, let us define the rate parameter as a function of another stochastic process to represent how we believe the rate is supposed to be distributed over the population of agents. Putting this together with the simple model, we now allow for our agents to be heterogeneous with regard to the rates at which they perform such an action. This is one of the fundamental aspects of why probabilistic modeling can provide amazingly accurate predictions when properly applied. Given that we used another stochastic process to bring in heterogeneity into the rate parameter, can you think of an analogous way of doing something similar with covariates?
Consider this: instead of telling the story that the rate parameter is spread over a population governed by a stochastic process, why don't we say that the rate parameter begins at a default rate and moves up or down in accordance to certain variables values of each particular agent? Use regression analysis to conjure up the relevant betas and then plug those into a function to modify the rate parameter. The great thing about using this method is that it does not necessarily prevent you from bringing in a stochastic process for the rate parameter itself. It is possible to implement both systems in tandem. Of course, with more complexity, you sacrifice more degrees of freedom, but in most cases, "over-fitting" will not be a problem if the number of covariates is kept under control.
I'm going to play around with this and probably put something up in the WIPs section when the initial write-up and tests are done.
1 Comments:
-
Will Dwinnell said...
-
"...in most cases, "over-fitting" will not be a problem if the number of covariates is kept under control."
That is not necessarily true. Even a very small number of variables may allow over-fitting, depending on the nature of the data and the modeling process employed. Beyond that, there is the issue of how to select even a small number of variables, in the event that many are available. For instance, if exactly 3 variables are chosen from 20 variables, there are over 1,100 possibilities- alot of room for over-fitting.
Monday, February 20, 2006
Killing 'E' Part 1 (or the baloney that is data mining)
This feels like something only a MGMT concentrator would do. Please humor me, will you? After you are done with that one, follow up.
In honor or President's Day, the U.S. markets have halted trading for a day. There are no indicators coming out today either, so let's just cut to the chase.
Quite by accident, I came across more information about the analysis technique called data mining through an unpublished manuscript under review. The paper itself was a complete mess, but from reading it, I could not help but think that there was simply something fundamentally amiss with the entire method. The concept is simple. Based on demographic information collected, data mining seeks to build profiles or come up with predictive variables that help firms determine market segments and CLV amongst other things. While I have no extensive background in the discipline, I quickly found fault with the approach.
Data mining to put it bluntly is regression analysis on a powerful anabolic steroid like D-bol. Interestingly enough, D-bol was banned by the FDA just about the time when data mining started becoming popular and feasible. Coincidence? Anyway, data mining techniques take quantitative and qualitative data and seek to find correlations between the data and the variable of interest such as customer purchases of a product. The data can range from income level to geographic location to age and gender.
The problem starts with the data preprocessing and the discretization of data. Ostensibly used to improve predicative ability, discretization takes a continuously valued attribute like personal
income and transforms it into discrete range values such as low (1), medium (2), and high (3). The discretized results are then used to represent the original values. Am I being overly critical here when I point out, whenever you aggregate data in this way, you will almost always get a better prediction no matter how arbitrarily you set up the discrete cutoffs. Preprocessing improves predictive capability precisely because it destroys variance in the raw data. It destroys 'E,' the random error term regression people hate. I admit, one may be able to determine cutoffs at which prediction capabilities are greater than aggregating the entire dataset into a binary value such as "Has income? Y/N," but then there is the question of what is so significant about these cutoffs other than the fact that they maximize predictive power.
At the heart of any model, there has to be a sensical story that can be told about the results of the analysis. More often than not, data mining tells ex post, cock-a-mammy stories. What is the reason behind the fact your high income bracket purchases more soy milk and why does it happen above the $77,326 annual income mark? Data miners can't tell you that, because before they ran the programs, they had no idea that this was going to be the output. They only discovered the fact ex post and had no tenable reason ex ante of what to expect.
Killing 'E' will almost always give you better answers. But then, so does making up data.
In honor or President's Day, the U.S. markets have halted trading for a day. There are no indicators coming out today either, so let's just cut to the chase.
Quite by accident, I came across more information about the analysis technique called data mining through an unpublished manuscript under review. The paper itself was a complete mess, but from reading it, I could not help but think that there was simply something fundamentally amiss with the entire method. The concept is simple. Based on demographic information collected, data mining seeks to build profiles or come up with predictive variables that help firms determine market segments and CLV amongst other things. While I have no extensive background in the discipline, I quickly found fault with the approach.
Data mining to put it bluntly is regression analysis on a powerful anabolic steroid like D-bol. Interestingly enough, D-bol was banned by the FDA just about the time when data mining started becoming popular and feasible. Coincidence? Anyway, data mining techniques take quantitative and qualitative data and seek to find correlations between the data and the variable of interest such as customer purchases of a product. The data can range from income level to geographic location to age and gender.
The problem starts with the data preprocessing and the discretization of data. Ostensibly used to improve predicative ability, discretization takes a continuously valued attribute like personal
income and transforms it into discrete range values such as low (1), medium (2), and high (3). The discretized results are then used to represent the original values. Am I being overly critical here when I point out, whenever you aggregate data in this way, you will almost always get a better prediction no matter how arbitrarily you set up the discrete cutoffs. Preprocessing improves predictive capability precisely because it destroys variance in the raw data. It destroys 'E,' the random error term regression people hate. I admit, one may be able to determine cutoffs at which prediction capabilities are greater than aggregating the entire dataset into a binary value such as "Has income? Y/N," but then there is the question of what is so significant about these cutoffs other than the fact that they maximize predictive power.
At the heart of any model, there has to be a sensical story that can be told about the results of the analysis. More often than not, data mining tells ex post, cock-a-mammy stories. What is the reason behind the fact your high income bracket purchases more soy milk and why does it happen above the $77,326 annual income mark? Data miners can't tell you that, because before they ran the programs, they had no idea that this was going to be the output. They only discovered the fact ex post and had no tenable reason ex ante of what to expect.
Killing 'E' will almost always give you better answers. But then, so does making up data.
Saturday, February 18, 2006
A brief respite (or nothing too special today)
This feels like something only a MGMT concentrator would do. Please humor me, will you? After you are done with that one, follow up.
The PPI released for January came in at 0.3%. Ex-food and energy the growth in producer prices was 0.4%. These are prices in the finished good category. The market tracks the finished goods index most closely, as it represents prices for goods that are ready for sale to the end user. In general, there are three categories within the PPI: crude, intermediate, and finished goods. The PPI for intermediate goods rose by 1.2% in January suggesting that there may be some more inflation coming down the line. If firms are unable to pass on that increased cost to consumers, the threat of inflation lowers. Since growth in prices in intermediate goods exceed finished goods, this seems to be the case at the moment. Still, increased prices for firms means a decrease in profit so it is definitely not desirable.
The PPI released for January came in at 0.3%. Ex-food and energy the growth in producer prices was 0.4%. These are prices in the finished good category. The market tracks the finished goods index most closely, as it represents prices for goods that are ready for sale to the end user. In general, there are three categories within the PPI: crude, intermediate, and finished goods. The PPI for intermediate goods rose by 1.2% in January suggesting that there may be some more inflation coming down the line. If firms are unable to pass on that increased cost to consumers, the threat of inflation lowers. Since growth in prices in intermediate goods exceed finished goods, this seems to be the case at the moment. Still, increased prices for firms means a decrease in profit so it is definitely not desirable.
Friday, February 17, 2006
The Bernanke Effect Part 3 (getting old already?)
This feels like something only a MGMT concentrator would do. Please humor me, will you? After you are done with that one, follow up.
Housing starts surged 14.5% month on month in January. The warm weather (global warming?) is a positive productivity shock, allowing for more construction. However, this is likely to be temporary, since house sales are still weakening. What may end up happening is that many of these houses go into inventory implying a slowdown in production later.
Unemployment claims continue to drop, suggesting that the labor market continues to strengthen going forward. Whether the decline in unemployment is inflationary remains to be seen. It will depend on factors such as cost of unit labor.
The second round of Ben Bernanke's testimony is before the Senate. The focus is definitely on trade and China. He noted that a 6% CA deficit is not sustainable indefinitely, but it is hard to say how long it will be sustained and how long it will take to reverse. Remember that the CA deficit is the flip side of the trade deficit NX. As the U.S. continues to borrow in order to consume, foreign creditors will be less and less likely to lend. There will come a time when foreign lender may decide to diversify away from dollar denominated assets, which may send the value of those assets plunging should there be a sell-off. There were lots of questions about the deficit with China, but Ben generally disagrees with tariffs (smart), since they restrict the gains from trade. Also, tariffs without changes in fundamentals such as increased savings rates will only result in shifting the balance of imports to other countries and not really helping the CA or NX deficits overall.
Housing starts surged 14.5% month on month in January. The warm weather (global warming?) is a positive productivity shock, allowing for more construction. However, this is likely to be temporary, since house sales are still weakening. What may end up happening is that many of these houses go into inventory implying a slowdown in production later.
Unemployment claims continue to drop, suggesting that the labor market continues to strengthen going forward. Whether the decline in unemployment is inflationary remains to be seen. It will depend on factors such as cost of unit labor.
The second round of Ben Bernanke's testimony is before the Senate. The focus is definitely on trade and China. He noted that a 6% CA deficit is not sustainable indefinitely, but it is hard to say how long it will be sustained and how long it will take to reverse. Remember that the CA deficit is the flip side of the trade deficit NX. As the U.S. continues to borrow in order to consume, foreign creditors will be less and less likely to lend. There will come a time when foreign lender may decide to diversify away from dollar denominated assets, which may send the value of those assets plunging should there be a sell-off. There were lots of questions about the deficit with China, but Ben generally disagrees with tariffs (smart), since they restrict the gains from trade. Also, tariffs without changes in fundamentals such as increased savings rates will only result in shifting the balance of imports to other countries and not really helping the CA or NX deficits overall.
Thursday, February 16, 2006
The Bernanke Effect Part 2 (Bravo, Ben)
The verdict is out: the markets liked Bernanke's testimony. The Dow is at its highest level since June 2001, even before 911. Oil prices dropped 3% to $57, reducing inflationary pressures on the economy. The testimony also included the Fed's forecasts for the next two years. Growth of real GDP is predicted to be at 3.5% in 2006 and 3-3.5% in 2007. Inflation will be around 2% in 2006 and 1.75-2% in 2007. Unemployment fluctuate between 4.75-5% in 2006 and '07.
Higher growth with less inflation is great for stocks. Lower inflation is great for bonds; however, there is still a large possibility that the Fed will raise rates once again in March and at subsequent meetings. In the short run, uncertainty is still high.
For his first time, Bernanke did a wonderful job. His speech was clear and he deftly avoided commenting on sensitive policy issues unlike Greenspan. For example, regarding the role of tax cuts in increasing economic inequality, he replied that where and how to make the tradeoff is a value judgment for Congress to make. It has no direct connection to monetary policy. However, he did say that it is important to reduce the deficit and deal with the social security problem, and that there are more efficient ways to help low-income workers than to raise the minimum wage.
A look at some other things he mentioned in his testimony.
Regarding the housing market Bernanke pointed to the slowing growth of home equity, which may lead households to increase savings and cut back on consumption expenditure. To a certain extent, the growth in household wealth and low interest rates have led to the low level of private savings. In the foreseeable future, savings will likely increase from its current level. Looking at other indicators surrounding the housing market, we can see declines in new housing starts and falling mortgages rates. Mortgage rates average at 7.9%, the lowest since December 2003 and are down 26% since their peak last June.
Bernanke noted that profits in the business sector continue to grow last year, boosted in part by continuing advances in productivity. Looking at the most recent industrial production numbers, we can see that while industrial production dropped by .2% in January, a significant portion of it was attributed to a 10% decline in utilities output due to the warm weather. Ex-utilities, factory output rose by .7%. Car parts and business equipment rose, perhaps signaling in increase in business investment. Businesses are increasing investment spending.
Higher growth with less inflation is great for stocks. Lower inflation is great for bonds; however, there is still a large possibility that the Fed will raise rates once again in March and at subsequent meetings. In the short run, uncertainty is still high.
For his first time, Bernanke did a wonderful job. His speech was clear and he deftly avoided commenting on sensitive policy issues unlike Greenspan. For example, regarding the role of tax cuts in increasing economic inequality, he replied that where and how to make the tradeoff is a value judgment for Congress to make. It has no direct connection to monetary policy. However, he did say that it is important to reduce the deficit and deal with the social security problem, and that there are more efficient ways to help low-income workers than to raise the minimum wage.
A look at some other things he mentioned in his testimony.
Regarding the housing market Bernanke pointed to the slowing growth of home equity, which may lead households to increase savings and cut back on consumption expenditure. To a certain extent, the growth in household wealth and low interest rates have led to the low level of private savings. In the foreseeable future, savings will likely increase from its current level. Looking at other indicators surrounding the housing market, we can see declines in new housing starts and falling mortgages rates. Mortgage rates average at 7.9%, the lowest since December 2003 and are down 26% since their peak last June.
Bernanke noted that profits in the business sector continue to grow last year, boosted in part by continuing advances in productivity. Looking at the most recent industrial production numbers, we can see that while industrial production dropped by .2% in January, a significant portion of it was attributed to a 10% decline in utilities output due to the warm weather. Ex-utilities, factory output rose by .7%. Car parts and business equipment rose, perhaps signaling in increase in business investment. Businesses are increasing investment spending.
Wednesday, February 15, 2006
The Bernanke Effect (Reflections on the new Fed Chair)
This feels like something only a MGMT concentrator would do. Please humor me, will you? After you are done with that one, follow up.
Retail sales were up a surprising 2.3% month over month in January. Analyst say increases in giftcard usage contributed to these positive numbers. The good news is that the gains went beyond autos. Excluding auto sales, retail sales were up 1.8%. As a result analysts have updated their estimates for Quarter1 GDP to about 4-5%. Remember, retail sales does not include purchases of services which account for more than half of all consumer spending. Despite that, it still remains a good leading indicator of what total consumption should be.
There were other good things happening as oil prices dropped below $60/barrel, the high price of oil ultimately being unsustainable. Over the last two days the Dow surged about 1.5% with the good news. Of course with the increase in retail sales, bond prices dropped after the report came out due to increased inflationary pressures.
Now about the new Fed Chair, Ben Bernanke.
Dr. Martin Asher, one of the first professors who taught me here at Penn, long predicted the succession of Alan Greenspan by Ben. After the announcement last year, he would say to his students when the subject was broached, "See, I that's what I've been saying would happen." Which got me to think what else would he know?
Dr. Jeremy Siegel, the man of finance here at Wharton, provides these descriptives about Ben: 1380 SAT, summa cum laude from Harvard, and National Spelling Bee finalist. Is it possible to discern the type of Fed Chair Ben would be? Let's try reading these tea leaves.
Truth be told, Ben is definitely taking over the role of Fed Chair at an uncertain time. Alan Greenspan leaves the post at the end of 12 consecutive rate hikes, bringing the funds rate to 4.5%. It was a no-brainer for Greenspan. As the U.S. economy recovered and gained traction, the interest rate hikes were set into action to slow down the economic and keep inflation under control. Going forward, the question is what should Ben do? Is it time to tighten the spigot some more or are we done with rate hikes? It sounds like a simple enough decision to make: rate up, rate down. If only! This is probably one of the hardest binary decisions to make on Earth.
In any case, we will probably be stuck with Ben for a long, long time. Alan Greenspan was forced to resign after 18.5 years as the Fed Chair due to term limits set by Federal law. Ben, if he serves out his maximum time, will be there for 28 years to retire at age 79 just like AG. This is guy everybody has got to live with.
Going foward: inflation, Fisher's equation, liquidity trap, depression...
...Is it for the children? Is it for himself? Is it for a memory? FFVII:AC (go watch it now)
Retail sales were up a surprising 2.3% month over month in January. Analyst say increases in giftcard usage contributed to these positive numbers. The good news is that the gains went beyond autos. Excluding auto sales, retail sales were up 1.8%. As a result analysts have updated their estimates for Quarter1 GDP to about 4-5%. Remember, retail sales does not include purchases of services which account for more than half of all consumer spending. Despite that, it still remains a good leading indicator of what total consumption should be.
There were other good things happening as oil prices dropped below $60/barrel, the high price of oil ultimately being unsustainable. Over the last two days the Dow surged about 1.5% with the good news. Of course with the increase in retail sales, bond prices dropped after the report came out due to increased inflationary pressures.
Now about the new Fed Chair, Ben Bernanke.
Dr. Martin Asher, one of the first professors who taught me here at Penn, long predicted the succession of Alan Greenspan by Ben. After the announcement last year, he would say to his students when the subject was broached, "See, I that's what I've been saying would happen." Which got me to think what else would he know?
Dr. Jeremy Siegel, the man of finance here at Wharton, provides these descriptives about Ben: 1380 SAT, summa cum laude from Harvard, and National Spelling Bee finalist. Is it possible to discern the type of Fed Chair Ben would be? Let's try reading these tea leaves.
Truth be told, Ben is definitely taking over the role of Fed Chair at an uncertain time. Alan Greenspan leaves the post at the end of 12 consecutive rate hikes, bringing the funds rate to 4.5%. It was a no-brainer for Greenspan. As the U.S. economy recovered and gained traction, the interest rate hikes were set into action to slow down the economic and keep inflation under control. Going forward, the question is what should Ben do? Is it time to tighten the spigot some more or are we done with rate hikes? It sounds like a simple enough decision to make: rate up, rate down. If only! This is probably one of the hardest binary decisions to make on Earth.
In any case, we will probably be stuck with Ben for a long, long time. Alan Greenspan was forced to resign after 18.5 years as the Fed Chair due to term limits set by Federal law. Ben, if he serves out his maximum time, will be there for 28 years to retire at age 79 just like AG. This is guy everybody has got to live with.
Going foward: inflation, Fisher's equation, liquidity trap, depression...
...Is it for the children? Is it for himself? Is it for a memory? FFVII:AC (go watch it now)
Tuesday, February 14, 2006
Love, at what cost? (let me just answer this one now: $113)
Legend says it started with a letter from a prisoner, Valentine, to his beloved - his jailer's daughter. And now, every February 14, gifts and cards are exchanged between sweethearts. But as more and more friends and family began exchanging gifts as well, Valentine's Day turned into a major retail holiday in the US, about four years or so ago. The spending continues to steadily grow, and this year the average cost of showing your love is $113 according to Brand Keys Inc., a boutique marketing consulting firm in New York City.
Once a holiday for lovers and schoolchildren, Valentine's Day is has expanded to include exchanges of cards and other tokens of affection between girlfriends, parents and children, and extended family members. Here is some other interesting information.
1. Consumers will be spending more than $1.7 billion dollars on candy, making Valentine's Day fourth in the holiday candy sweepstakes behind Halloween, Easter, and Christmas.
2. Valentine's Day is the second-largest greeting-card holiday, just behind Christmas, with estimated Valentine card sales of $2.3 billion in 2006.
3. Valentine's Day continues to account for nearly 33% of holiday sales of all cut flowers, more than any other holiday.
The U.S. has really to the point where we think a really expensive gift may be better than a personally written sonnet (well, there's nothing stopping you to write that sonnet and take her out to dinner). What a pity. If you must splurge, think of the other things you could be doing with $113. Some suggestions:
1. Import a set of bonsai tree pruning tools, from Amazon Japan.
2. Vitamins. A lot of 'em. Call it the gift of life.
3. An electric stapler. One of my favorites. The kind you press yourself is so last century.
Happy Valentines Day.
Monday, February 13, 2006
Savings, is there money in you pocket? (on the current account deficit)
Tomorrow, the retail sales figures will be coming out at 8:30am. It's a very important leading indicator of consumer spending patterns. We can't get the whole picture on consumer spending based on retail sales because the figure does not include spending on services, which makes up over half of total consumption. Retail sales are also given ex-auto purchases as those prove to be very volitile from month to month.
Last week, I wrote a little about the trade deficit and I mentioned how the flow of real goods like computers, cars, and sneakers are inimately related to the flow of money. Foreign lenders lend their wealth to the U.S. as measured by the current account (CA). For the United States, the trade deficit (net exports, NX) and the current account at flips sides of each other.
Remember the accounting function for GDP: Y = C + I + G + NX. Y here is GDP. C is consumpution. I is investment. G is government spending, and NX is net exports.
Rearranging the formula, we can find that NX = Y - C - I - G. Let us define Savings (S) as S = Y - C - G. Plug savings into the formula for net exports and we get NX = S - I.
What the last equation says is that the value of net exports is equal to the level of domestic savings minus domestic investment. When domestic investment is greater than savings, we run a trade deficit. Remember from last time I said NX ~= CA. Following the transaction between real goods and money, we can also write CA = S - I.
Investment growth has largely remained steady historically, but savings in the U.S. has dropped to lower and lower levels since the 1990s. This relationship is at the core of the trade deficit problem. Part of the reason for the low savings rate in the U.S. is due to the large government budget deficits that the administration runs. Last week, I alluded to the relation between budget and trade deficits. When the government overspends, it must borrow money with bonds and other financial products. Government savings is then a negative number. This widens the gap between S and I resulting in higher CA and NX's. More on this topic in the future.
Last week, I wrote a little about the trade deficit and I mentioned how the flow of real goods like computers, cars, and sneakers are inimately related to the flow of money. Foreign lenders lend their wealth to the U.S. as measured by the current account (CA). For the United States, the trade deficit (net exports, NX) and the current account at flips sides of each other.
Remember the accounting function for GDP: Y = C + I + G + NX. Y here is GDP. C is consumpution. I is investment. G is government spending, and NX is net exports.
Rearranging the formula, we can find that NX = Y - C - I - G. Let us define Savings (S) as S = Y - C - G. Plug savings into the formula for net exports and we get NX = S - I.
What the last equation says is that the value of net exports is equal to the level of domestic savings minus domestic investment. When domestic investment is greater than savings, we run a trade deficit. Remember from last time I said NX ~= CA. Following the transaction between real goods and money, we can also write CA = S - I.
Investment growth has largely remained steady historically, but savings in the U.S. has dropped to lower and lower levels since the 1990s. This relationship is at the core of the trade deficit problem. Part of the reason for the low savings rate in the U.S. is due to the large government budget deficits that the administration runs. Last week, I alluded to the relation between budget and trade deficits. When the government overspends, it must borrow money with bonds and other financial products. Government savings is then a negative number. This widens the gap between S and I resulting in higher CA and NX's. More on this topic in the future.
Saturday, February 11, 2006
The positive and the normative (or why there is hope for the 'lapsed')
On Friday, the December trade balance figures were released by the Department of Commerce at 8:30am. It was a whopping -$65.7 bn. This was greater (more negative) than the market expectation of -$64.8 bn. This brings the cumulative trade deficit for the U.S. in 2005 to $725.8 bn, roughly 6% of GDP. Of course having the trade deficit hit an all time low has set off another round of finger pointing. The most obvious culprit is China, who accounts for $201.6 bn or 28% of the deficit, but there were other factors that contributed to the growth of the trade deficit. Natural disasters like Hurricane Katrina knocked out oil production on the Gulf coast leading to more foreign oil imports.
When goods come into the U.S., cash (cash equivalents) goes out to pay for these purchases. The exporting countries then take this money and buys up U.S. dollars, stocks, bonds and other investments. Therefore, a duality exists between financial markets and markets for real goods. They are flip sides of each other. The trouble with foreign ownership of U.S. assets according to economists is that if they suddenly decide to hold fewer U.S. assets, foreigners could send the value of the dollar, stocks and bonds all plunging.
Well, what should be done? Tariffs? Protectionism? No, that would only shift the distribution of imports from China to elsewhere, and it would put inflationary pressure on the economy due to the increases in prices. This is where macroeconomists begin telling the story of the Twin Deficits: Trade and Budget. I think I'll save this story for Monday, because otherwise I'll never get to the topic for today. Watch out for the retail sales numbers coming out on Tuesday at 830 sharp!
The positive and the normative
The story is as follows: "A child is born and raised in a devoutly religious family. She grows up and goes to college. While in college she is exposed to ideas that cause her to begin to question the faith she was taught to believe. As a result, she rejects the religion and joins the ranks of the 'lapsed' - a person raised in a religion but no longer practices it. This is a story that plays itself out again and again across the U.S.
The story at its heart is about the conflict of two value systems. There is the religious value system that instructs us to believe, and there is the scientific value system that teaches us to doubt. Finding it impossible to serve these two masters, many find themselves accepting the latter system and outright rejecting the former. I wonder, being a Catholic myself, if these two ideas are truly irreconcilable with each other.
Science or 'the doubting way' is a practice by which we are able to understand natural phenomenon. We apply its methods to test things and posit causal relationships. The offshoot of science, technology, is the application of our knowledge of the world. Yet, neither science nor technology provides its practitioners with a set of ethics. They are both things that make positive statements but give us no normative prescription. So, it is possible to do great good or great evil with science and technology. Flight has given us the convenience of air travel but has also brought to us the horrors of aerial warfare. Nuclear power has the potential to provide us with an unlimited supply of energy, but it has also brought upon humanity the scepter of instant and total annihilation. In the wake of our discoveries, there is an ethical void which cannot be filled with science. What about religion?
The problem with religion is that it tries to straddle two boats. On one hand it tries to make claims about the physical world. On the other, it tries to provide ethical guidance to its adherents. While there are groups of people who believe that theology can lead us to better understanding the physical world, the majority of us have come to reject that part of religion. What is valuable about religion is that it can serve as an ethical blueprint. Of course, religion is not alone in this field as there are countless other value propositions that can be used for ethical guidance. In a debating society, these various systems can and should be brought into the open and compared (and they are, looking at some of the most contentious issues of the day such as the pro-life and pro-choice debate).
Maybe the 'lapsed' missed an important point when they rejected their religion. Maybe despite all our progress in science, religion still holds an indispensable place in society. Science can tell us 'why,' but who will tell us 'how' and 'should'?
Snow has finally arrived in Philadelphia with a vengence.
When goods come into the U.S., cash (cash equivalents) goes out to pay for these purchases. The exporting countries then take this money and buys up U.S. dollars, stocks, bonds and other investments. Therefore, a duality exists between financial markets and markets for real goods. They are flip sides of each other. The trouble with foreign ownership of U.S. assets according to economists is that if they suddenly decide to hold fewer U.S. assets, foreigners could send the value of the dollar, stocks and bonds all plunging.
Well, what should be done? Tariffs? Protectionism? No, that would only shift the distribution of imports from China to elsewhere, and it would put inflationary pressure on the economy due to the increases in prices. This is where macroeconomists begin telling the story of the Twin Deficits: Trade and Budget. I think I'll save this story for Monday, because otherwise I'll never get to the topic for today. Watch out for the retail sales numbers coming out on Tuesday at 830 sharp!
The positive and the normative
The story is as follows: "A child is born and raised in a devoutly religious family. She grows up and goes to college. While in college she is exposed to ideas that cause her to begin to question the faith she was taught to believe. As a result, she rejects the religion and joins the ranks of the 'lapsed' - a person raised in a religion but no longer practices it. This is a story that plays itself out again and again across the U.S.
The story at its heart is about the conflict of two value systems. There is the religious value system that instructs us to believe, and there is the scientific value system that teaches us to doubt. Finding it impossible to serve these two masters, many find themselves accepting the latter system and outright rejecting the former. I wonder, being a Catholic myself, if these two ideas are truly irreconcilable with each other.
Science or 'the doubting way' is a practice by which we are able to understand natural phenomenon. We apply its methods to test things and posit causal relationships. The offshoot of science, technology, is the application of our knowledge of the world. Yet, neither science nor technology provides its practitioners with a set of ethics. They are both things that make positive statements but give us no normative prescription. So, it is possible to do great good or great evil with science and technology. Flight has given us the convenience of air travel but has also brought to us the horrors of aerial warfare. Nuclear power has the potential to provide us with an unlimited supply of energy, but it has also brought upon humanity the scepter of instant and total annihilation. In the wake of our discoveries, there is an ethical void which cannot be filled with science. What about religion?
The problem with religion is that it tries to straddle two boats. On one hand it tries to make claims about the physical world. On the other, it tries to provide ethical guidance to its adherents. While there are groups of people who believe that theology can lead us to better understanding the physical world, the majority of us have come to reject that part of religion. What is valuable about religion is that it can serve as an ethical blueprint. Of course, religion is not alone in this field as there are countless other value propositions that can be used for ethical guidance. In a debating society, these various systems can and should be brought into the open and compared (and they are, looking at some of the most contentious issues of the day such as the pro-life and pro-choice debate).
Maybe the 'lapsed' missed an important point when they rejected their religion. Maybe despite all our progress in science, religion still holds an indispensable place in society. Science can tell us 'why,' but who will tell us 'how' and 'should'?
Snow has finally arrived in Philadelphia with a vengence.
Friday, February 10, 2006
Process utility Part 1 (or how to avoid existential angst)
Beware, the following writing is less sober and waxes more philosophical in a perversely, economic way than my usual writing. I am even writing in sentence fragments! It is a very interesting post though, and I recommend you read it.
So there's this guy, John Paul Sartre, who once said: "Every existing thing is born without reason, prolongs itself out of weakness, and dies by chance." I may not have gotten the wording down exactly, but you get the idea. Adam Smith wrote in the Theory of Moral Sentiments something to the effect that our overestimation of the utility from future wealth is what drives our economy, but once we get to the future, we will suddenly realize what a folly it was to spend so much of our life and effort pursuing this "trinkets" of vanity. Trinkets of limited utility (frag). So it is with fame and all things ephemeral (frag). On the other hand, if everyone accepted these ideas put forth by Smith and Sartre, what will become of this world?
The fact that humans are alive on this earth means that we each must confront the purpose of existence. A person who has limited capacity to ponder this issue is fortunate. They are shielded from being torn apart by the meaning of being. I think the burden falls far more heavily on those who have the time and capacity to confront this question. The rational mind languishes, agonizes without respite except when it find itself anxious over other urgent things that calls out for its attention. Then it may be distracted for a moment, perhaps by intention, perhaps without knowing it, but it will eventually meander its way back. It is the curse of the intelligence, and it only magnifies itself the greater one's ability becomes.
Someone with great abilities who has not settled on this issue is the worst sort, because they do not know why they struggle. In their distress they continue to bury themselves in their labors without understanding why is it that they never feel satisfaction, or why do they always feel inferior when their accomplishments have already far outshone the nearest competitor. Thus they sink into hopelessness. Nothing they do can ever be enough. However, that begs the question: "enough for what?"
What should we do? Is there something we can do? Could we do it? I think there is a way.
The fortress of classical economics built on Walrasian precepts is being torn down stone by stone these last thirty years. Academics have realized what Adam Smith has known all along. People are not merely concerned with outcomes. There is inherent value in what we perceive as procedural justice and procedural fairness. Outcomes are not the only things human beings care about. Humor me by considering this example.
It is true that both the vagrant and the rich man leave this world not being able to take anything with them. I posit that their outcomes (death) lie on the same indifference curve. Should I be partial to the rich man and not the vagrant? Who cares? They are both dead! This choice is uninteresting. Asking the question: "Who's life would I have wanted to live?" elicits a far more interesting response. Well, I hate to be wishy-washy, but like an economist, I am bound to say "It depends. It depends. Really." I leave it to you to work out the conditions under which one may be preferable to the other and vice versa. Perhaps going through this simple exercise may help you discover what are the things that you truly value. In this world where the ultimate outcome does not matter, we must care about procedure.
Religion presents an attractive alternative because it makes a promise as to the outcomes: life after death, reincarnation, heaven, hell. Unfortunately, for some people, they find this answer unsatisfying. "We cannot be sure." "We do not believe." "We refuse to accept or place our trust in these fantastic stories." Blah blah blah. Rational thinkers often struggle with religion (although Pascal will argue it's perfectly rational to believe.). For someone boundedly rational, there is still hope that they may find their deliverance in the embrace of religion, but even that is not a guarantee.
The fact that we am alive here on this earth and that outcomes do not matter ultimately should not only be seen as a dismal limiter. In fact, it is a empowering concept. In life you seldom get the chance to struggle with everything on the line. If I accept the fact that my ultimate outcome is death, then I should have no inhibition to struggle. It would be a great adventure. For my dream/goal/ambition, I am willing to put it all on the line, because I only get this chance. I cannot be sure that I will return after I am gone; therefore, given the opportunity, I can do no worse.
Process may not be a means to an end, because process utility matters. In a world where we cannot perceive our ultimate outcomes (beyond death), all we have is process. Luckily for us, process is something we can care about.
So there's this guy, John Paul Sartre, who once said: "Every existing thing is born without reason, prolongs itself out of weakness, and dies by chance." I may not have gotten the wording down exactly, but you get the idea. Adam Smith wrote in the Theory of Moral Sentiments something to the effect that our overestimation of the utility from future wealth is what drives our economy, but once we get to the future, we will suddenly realize what a folly it was to spend so much of our life and effort pursuing this "trinkets" of vanity. Trinkets of limited utility (frag). So it is with fame and all things ephemeral (frag). On the other hand, if everyone accepted these ideas put forth by Smith and Sartre, what will become of this world?
The fact that humans are alive on this earth means that we each must confront the purpose of existence. A person who has limited capacity to ponder this issue is fortunate. They are shielded from being torn apart by the meaning of being. I think the burden falls far more heavily on those who have the time and capacity to confront this question. The rational mind languishes, agonizes without respite except when it find itself anxious over other urgent things that calls out for its attention. Then it may be distracted for a moment, perhaps by intention, perhaps without knowing it, but it will eventually meander its way back. It is the curse of the intelligence, and it only magnifies itself the greater one's ability becomes.
Someone with great abilities who has not settled on this issue is the worst sort, because they do not know why they struggle. In their distress they continue to bury themselves in their labors without understanding why is it that they never feel satisfaction, or why do they always feel inferior when their accomplishments have already far outshone the nearest competitor. Thus they sink into hopelessness. Nothing they do can ever be enough. However, that begs the question: "enough for what?"
What should we do? Is there something we can do? Could we do it? I think there is a way.
The fortress of classical economics built on Walrasian precepts is being torn down stone by stone these last thirty years. Academics have realized what Adam Smith has known all along. People are not merely concerned with outcomes. There is inherent value in what we perceive as procedural justice and procedural fairness. Outcomes are not the only things human beings care about. Humor me by considering this example.
It is true that both the vagrant and the rich man leave this world not being able to take anything with them. I posit that their outcomes (death) lie on the same indifference curve. Should I be partial to the rich man and not the vagrant? Who cares? They are both dead! This choice is uninteresting. Asking the question: "Who's life would I have wanted to live?" elicits a far more interesting response. Well, I hate to be wishy-washy, but like an economist, I am bound to say "It depends. It depends. Really." I leave it to you to work out the conditions under which one may be preferable to the other and vice versa. Perhaps going through this simple exercise may help you discover what are the things that you truly value. In this world where the ultimate outcome does not matter, we must care about procedure.
Religion presents an attractive alternative because it makes a promise as to the outcomes: life after death, reincarnation, heaven, hell. Unfortunately, for some people, they find this answer unsatisfying. "We cannot be sure." "We do not believe." "We refuse to accept or place our trust in these fantastic stories." Blah blah blah. Rational thinkers often struggle with religion (although Pascal will argue it's perfectly rational to believe.). For someone boundedly rational, there is still hope that they may find their deliverance in the embrace of religion, but even that is not a guarantee.
The fact that we am alive here on this earth and that outcomes do not matter ultimately should not only be seen as a dismal limiter. In fact, it is a empowering concept. In life you seldom get the chance to struggle with everything on the line. If I accept the fact that my ultimate outcome is death, then I should have no inhibition to struggle. It would be a great adventure. For my dream/goal/ambition, I am willing to put it all on the line, because I only get this chance. I cannot be sure that I will return after I am gone; therefore, given the opportunity, I can do no worse.
Process may not be a means to an end, because process utility matters. In a world where we cannot perceive our ultimate outcomes (beyond death), all we have is process. Luckily for us, process is something we can care about.
Thursday, February 09, 2006
Post-feminism (or how boys are flunking Life 101)
(deleted)
Wednesday, February 08, 2006
Tolerating the intolerant (or why it is dangerous to suppress the intensely offensive)
On September 30, 2005, the Danish Jyllands-Posten newspaper published an article depicting the Islamic prophet Muhammad in 12 different cartoons. The reaction to these pictures, which were seen as blasphemous, from members of the Muslim community has been fierce, especially in Islamic countries. Death threats have been issued to people associated with the paper. Danish consulates have been attacked as well as Danish troops in Afghanistan. The intolerance and attempt by some members of the Muslim community to suppress expression has been most disturbing. Despite my disappointment with the Muslim reaction, I am even more put off by the round of political correctness coming from politicians who say, "We should respect freedom of speech, but not the kind that produces religiously offensive material." As spiritual inheritors of the Greek tradition of open forums, I had expected more than this weak-kneed response by leaders of the free world.
The central question to the raging debate is this: What are the limits to freedom of speech? Just asking it in this context begets other questions. How should people living in free societies deal with religious prejudice and live with people holding those beliefs? What should the law say in regards to what is tolerable and what is intolerable?
I am not an expert in law, but I do know science. Science, specifically Western science, is built on three things: truth, empirical methods, and open debate. Through the pursuit of truth using empirical methods in an open forum, Western science has showed humanity a path that is free from the veil of ignorance, superstition, and prejudice. I had in a previous post written extensively about the success of the scientific process as one which depends on uncertainty and the fact that new ideas are always welcome.
In the Western political tradition, the pursuit of truth is equally important as it is in science. To this end, Western democracy instituted open, public debate through free speech. If one group of people can use force or threat of force to propagate orthodoxy, then open debate dies. The death of open debate obscures truth, because new ideas will not be allowed to spread. The loss of truth leads to tyranny, be it tyranny from ignorance or tyranny from the keepers of orthodoxy.
You cannot suppress dissent through threat of arms. Dissent can be held in the mind. As China is painfully learning and will painfully learn in the future, the free flow of information through the internet provides dissent, the unorthodox, a way to spread like never before. To try to keep ideas contained is a losing battle. Let the ideas come out and be scrutinized in public and let the people be the judge of those ideas. The attempt to put a leash on every thought out there is an impossible task, and even if it is possible, the controllers can only claim a pyrrhic victory.
I applaud the stance held by the Danish Prime Minister Anders Fogh Rasmussen when he said, "The government refuses to apologize because the government does not control the media or a newspaper outlet; that would be in violation of the freedom of speech." That is to say, speech no matter how offensive, has the right to be expressed. I am sure there are legal limits to what can be said. To threaten the life of another for example obviously should not be protected by free speech, because the state is sworn to protect its individual members and because to threaten another with death brings an end to free speech. Outside of these limits, anything is game. How we express our views then is just a matter of our taste and upbringing.
To end, I think Martin Wolf in his commentary written for the FT today really sums it up when he writes, "Freedom is our birthright. Self-restraint is a measure of our maturity." Grow up, kids.
The central question to the raging debate is this: What are the limits to freedom of speech? Just asking it in this context begets other questions. How should people living in free societies deal with religious prejudice and live with people holding those beliefs? What should the law say in regards to what is tolerable and what is intolerable?
I am not an expert in law, but I do know science. Science, specifically Western science, is built on three things: truth, empirical methods, and open debate. Through the pursuit of truth using empirical methods in an open forum, Western science has showed humanity a path that is free from the veil of ignorance, superstition, and prejudice. I had in a previous post written extensively about the success of the scientific process as one which depends on uncertainty and the fact that new ideas are always welcome.
In the Western political tradition, the pursuit of truth is equally important as it is in science. To this end, Western democracy instituted open, public debate through free speech. If one group of people can use force or threat of force to propagate orthodoxy, then open debate dies. The death of open debate obscures truth, because new ideas will not be allowed to spread. The loss of truth leads to tyranny, be it tyranny from ignorance or tyranny from the keepers of orthodoxy.
You cannot suppress dissent through threat of arms. Dissent can be held in the mind. As China is painfully learning and will painfully learn in the future, the free flow of information through the internet provides dissent, the unorthodox, a way to spread like never before. To try to keep ideas contained is a losing battle. Let the ideas come out and be scrutinized in public and let the people be the judge of those ideas. The attempt to put a leash on every thought out there is an impossible task, and even if it is possible, the controllers can only claim a pyrrhic victory.
I applaud the stance held by the Danish Prime Minister Anders Fogh Rasmussen when he said, "The government refuses to apologize because the government does not control the media or a newspaper outlet; that would be in violation of the freedom of speech." That is to say, speech no matter how offensive, has the right to be expressed. I am sure there are legal limits to what can be said. To threaten the life of another for example obviously should not be protected by free speech, because the state is sworn to protect its individual members and because to threaten another with death brings an end to free speech. Outside of these limits, anything is game. How we express our views then is just a matter of our taste and upbringing.
To end, I think Martin Wolf in his commentary written for the FT today really sums it up when he writes, "Freedom is our birthright. Self-restraint is a measure of our maturity." Grow up, kids.
Tuesday, February 07, 2006
Power on the short side (or why even nasty customers can demand to be respected)
The motivating question for today's post is this: Why was it that American cars manufactured during the Cold War were so much better than their Russian counterparts?
Despite the troubles plaguing the American automobile industry today, there was a time in the past when the cars coming out of Detroit were unparalleled anywhere in the world. Okay, so that point is arguable, but for the sake of this discussion, let's take this to be fact by fiat. What was true during those times was that Russian manufacturing actually carried on as a value destroying process. Yes, raw materials going into the factories were actually more valuable than the products coming out.
Despite the troubles plaguing the American automobile industry today, there was a time in the past when the cars coming out of Detroit were unparalleled anywhere in the world. Okay, so that point is arguable, but for the sake of this discussion, let's take this to be fact by fiat. What was true during those times was that Russian manufacturing actually carried on as a value destroying process. Yes, raw materials going into the factories were actually more valuable than the products coming out.
What could be the reason behind this discrepancy? Was it because of intrinsic differences between Americans and Russians that the former's cars were better? Of course not, or at least the liberal part of me would argue vehemently that it is not. Maybe, just maybe the fact that Russians waited in line to get their Volgas and American salesmen* lined up to sell customers their Dodges, Fords, or Chevys made all the difference. Or putting it another way, there were people lined up to work at Ford's factory in Detroit.
Let's take a step back. In Econ 101, students are introduced to the classical models employed in economic analysis. Where the supply and demand curves intersect is where the market clears at a going price and given quantity. Aberrations occur when the prices are set above or below the price found at the intersection, resulting in shortage or surplus. Theoretically in a market economy, the equilibrium matches all willing buyers with willing suppliers. This is a state of equality, but looking at the real world, is the relationship between the two transacting agents really devoid of power?
Looking at just the two types of markets illustrated by the example, it can be seen that in both cases, there is one party that exercises some sort of power over the other. This power comes in the form of the power holder providing some sort of favor or more technically, an economic rent on the other. Why would the rentier be at the mercy of the rentee? This does not make sense, but if one thinks more carefully, it makes perfect sense. In both cases with the car salesmen and workers looking for jobs, they are at the mercy of the other party for two reasons. First, they are afraid of losing the rent they receive from the consumer or the employer. Second, there exists a group of people in both cases that are constrained and so do not receive the aforementioned rents. There are the salesmen who make a sale and there are those who don't. There are people who find employment and there are those who don't. Put the two ideas together. The party that receives the rent is afraid of losing it because there are people just waiting to snatch it from them if they lose it. This is what is called being on the long side.
The power relationship creates an incentive to do all that one can to retain the rent. Product and service quality rises under market systems. Workers technically work harder than they have to at their wage levels. Although it's not a perfect parallel, the fact that people queued up for rationed goods might be a reason for the notoriously inferior products produced by Communist countries. Those people had no other choice. The incentive structure under the American system is such that the quality and value of products will continue to increase over time. So the next time, you wonder why the customer service representatives are always so happy to hear your problems, you'll know why.
*pardon, there were only car salesmen in that age. I'll write a post about the post-feminist world we live in at some later date. Oooo... That'll be a good one.
Let's take a step back. In Econ 101, students are introduced to the classical models employed in economic analysis. Where the supply and demand curves intersect is where the market clears at a going price and given quantity. Aberrations occur when the prices are set above or below the price found at the intersection, resulting in shortage or surplus. Theoretically in a market economy, the equilibrium matches all willing buyers with willing suppliers. This is a state of equality, but looking at the real world, is the relationship between the two transacting agents really devoid of power?
Looking at just the two types of markets illustrated by the example, it can be seen that in both cases, there is one party that exercises some sort of power over the other. This power comes in the form of the power holder providing some sort of favor or more technically, an economic rent on the other. Why would the rentier be at the mercy of the rentee? This does not make sense, but if one thinks more carefully, it makes perfect sense. In both cases with the car salesmen and workers looking for jobs, they are at the mercy of the other party for two reasons. First, they are afraid of losing the rent they receive from the consumer or the employer. Second, there exists a group of people in both cases that are constrained and so do not receive the aforementioned rents. There are the salesmen who make a sale and there are those who don't. There are people who find employment and there are those who don't. Put the two ideas together. The party that receives the rent is afraid of losing it because there are people just waiting to snatch it from them if they lose it. This is what is called being on the long side.
The power relationship creates an incentive to do all that one can to retain the rent. Product and service quality rises under market systems. Workers technically work harder than they have to at their wage levels. Although it's not a perfect parallel, the fact that people queued up for rationed goods might be a reason for the notoriously inferior products produced by Communist countries. Those people had no other choice. The incentive structure under the American system is such that the quality and value of products will continue to increase over time. So the next time, you wonder why the customer service representatives are always so happy to hear your problems, you'll know why.
*pardon, there were only car salesmen in that age. I'll write a post about the post-feminist world we live in at some later date. Oooo... That'll be a good one.
Monday, February 06, 2006
Intrinsic differences (or why over a billion East Asians are making life harder for themselves)
There's no significant economic indicator coming out today or any at all this week for that matter. Maybe the trade balance report on Friday will be interesting to analyze, but that's Friday. It looks like the U.S. indicies will remain pretty much unchanged for the day. The portfolio is very much in the green.
Intrinsic differences
My entire love-hate relationship with Chinese orthography just grows and grows. On one hand, I respect it as an art form, but on the other hand, its lack all practicality. The more I study the script, the more I realize how it is so inferior to the alphabetic systems devised by the West. Mark my words, with the continuing development of high speed computing and the growth of computer dependency, the characters will fall further and further behind, and eventually die away. The people who use Chinese characters have the spend more time developing software for their scripts and they derive far less utility from it. Due to the intrinsic nature of script, it will never be able to catch up to the efficiency of the phonetic alphabets.
It takes 416 bits to encode all 52 capital and lower case letters of the English alphabet. It takes 80,920 bits to encode roughly the 6,400 Hanzi used in Japanese computing (and even this is only a limited set). Big5 requires slightly more than double that number coming in at around 13,000 Hanzi in the primary and secondary set. GB used in China is slightly smaller. At least the standard published in the 1980s promulgated the roughly 8,000 characters in the primary and secondary sets. Think: 2^16 gives us only 65,536 possible combinations. That's only 2 bytes. Chinese systems need at least 3 bytes for the encoding of the primary and secondary sets (and once again, neither set exhausts the number of characters out there!). How can Hanzi possibly compare against the streamlined efficiency of alphabetic script when it comes to computing? It can't! Recently, China has pushed a new standard, the GB 18030. Can you believe this: It's a 4 byte encoding scheme. 4 bytes! As if the entire system surrounding the characters is not clumsy enough already.
For more on this topic, go to the WIPs page and see the articles under the heading "On Chinese."
Intrinsic differences
My entire love-hate relationship with Chinese orthography just grows and grows. On one hand, I respect it as an art form, but on the other hand, its lack all practicality. The more I study the script, the more I realize how it is so inferior to the alphabetic systems devised by the West. Mark my words, with the continuing development of high speed computing and the growth of computer dependency, the characters will fall further and further behind, and eventually die away. The people who use Chinese characters have the spend more time developing software for their scripts and they derive far less utility from it. Due to the intrinsic nature of script, it will never be able to catch up to the efficiency of the phonetic alphabets.
It takes 416 bits to encode all 52 capital and lower case letters of the English alphabet. It takes 80,920 bits to encode roughly the 6,400 Hanzi used in Japanese computing (and even this is only a limited set). Big5 requires slightly more than double that number coming in at around 13,000 Hanzi in the primary and secondary set. GB used in China is slightly smaller. At least the standard published in the 1980s promulgated the roughly 8,000 characters in the primary and secondary sets. Think: 2^16 gives us only 65,536 possible combinations. That's only 2 bytes. Chinese systems need at least 3 bytes for the encoding of the primary and secondary sets (and once again, neither set exhausts the number of characters out there!). How can Hanzi possibly compare against the streamlined efficiency of alphabetic script when it comes to computing? It can't! Recently, China has pushed a new standard, the GB 18030. Can you believe this: It's a 4 byte encoding scheme. 4 bytes! As if the entire system surrounding the characters is not clumsy enough already.
For more on this topic, go to the WIPs page and see the articles under the heading "On Chinese."
Saturday, February 04, 2006
On proper methodology (or why an honest politician can never get elected)
Think back to seventh grade science. Recall the scientific method.
It is as follows:
1. Make an observation
2. Hypothesize a cause
3. Design a test of the hypothesis
4. Perform the test and collect the data
5. Evaluate your hypothesis against the data
6. Repeat
It's a process that's introduced to us as a disciplined, rational, and of course scientific approach to solving problems in the real world. We can take this method and apply it to all sorts of phenomenon, physical or not, and from it, arrive at well reasoned conclusions. The beauty of this method is that it is revealing but not conclusive. It's not only possible discover a proper way to think about observable phenomenon (observable here is used in the broadest sense in that something may be measured), but it is also possible to disprove established rules of causation should a new paradigm arise. There is nothing sacred, even the so called "laws of physics" are suspect to scrutiny. Although in their case, they have already been scrutinized so much that we can all safely agree we have them down with 99.999% confidence. Still, the point is that with this method, it is impossible to prove things for sure. All that can be said is that they have not yet been disproved at present, but they certainly may be in the future.
If that's the case, then why should the scientific method be used anyway? Isn't it desirable to know things for sure? It depends on who you ask. I can think of many cases in which people probably don't want to know for sure. In science, not knowing for sure is what motivates its practitioners to continue making new discoveries and re-testing old ideas. Ironically, in religion, it is probably not desirable to know for sure. People believe God exists, but what if one day someone comes along and claims that she knows God exists? If the basis for religion moves from being a hypothesis to being a more or less guaranteed fact, what will happen to religion? Even though having faith means that one believes the tenets established by the religion to be true, I do not think those with faith can face the knowledge that it is true. If it is true, then where is the value or redemption found in believing?
Is it ever good to know for sure? I cannot answer that with certainty (haha, lame joke here), but I can certainly think of situations in which certainty is desired over proper methodology. The example I am thinking of goes back to the heading of my entry: Why an honest politician can never get elected. Democratic politics today is awash with politicians who seem to know just the thing that the nation needs to fix its problems. They take to the road and campaign with promises of programs and initiatives that will solve all problems. All the people have to do is to trust them and elect them into office. Since anyone can make unsubstantiated claims, the game goes to the side who can woo the voters with promises of a rosier future. The result of this condition is that once in office, the elected become faced with the tough reality of governance and soon realize that the promises cannot be kept. Is there a way to fix this problem? Let's think about this for a minute.
Imagine two politicians running for some office. One says, "I know our district is going through some tough times right now. Many people don't have jobs. Crime runs rampant through our streets. Our children are not prepared to go out into the real world because our school system is broken. Times are rough, but I have a vision for our district. If you elect me into such and such an office, I promise to put into action initiatives that will reduce taxes, encourage business and job growth, strengthen our police force with the funding they need to fight crime, and fix our schools so that our children can enjoy a better future. Here is my plan: (plan here.)" When asked how he plans to deal with the rising costs of healthcare, he answers immediately, "I will do this, this, and this." The people are impressed.
The next day, the other politician running for the same office comes into the same town and addresses the same audience. He gives his speech addressing the problems similar to his opponent. During the Q & A, someone asks the same question about healthcare. Pausing to think about it for a moment, he then answers, "Well, I know the rising cost of healthcare is a growing concern for workers and employers in our district, and it's a very complicated issue. Honestly, I don't have a definite plan to solve it yet, but I will say that I will look into the matter very carefully. I promise to collect the input of all the people in this area and consult the experts. After that, my aides and I will put forth what we feel is the best plan for our district and have an open debate about its merits and shortcomings. If the plan is fundamentally sound and receives broad based support by the people, our government will put it into action. Afterwards, we will come back and evaluate the plan's performance and determine whether or not modifications need to be made." I pose this question: Who do you think will get elected?
Stop and think. Should the solution be more sacred than the process? I don't think honest politicians will ever be elected in this country.
On a personal note, it has been an unusually warm winter here in Philadelphia. I had promised the HK exchange student that the winters here were bitterly cold. He must be disappointed, and so am I.
It is as follows:
1. Make an observation
2. Hypothesize a cause
3. Design a test of the hypothesis
4. Perform the test and collect the data
5. Evaluate your hypothesis against the data
6. Repeat
It's a process that's introduced to us as a disciplined, rational, and of course scientific approach to solving problems in the real world. We can take this method and apply it to all sorts of phenomenon, physical or not, and from it, arrive at well reasoned conclusions. The beauty of this method is that it is revealing but not conclusive. It's not only possible discover a proper way to think about observable phenomenon (observable here is used in the broadest sense in that something may be measured), but it is also possible to disprove established rules of causation should a new paradigm arise. There is nothing sacred, even the so called "laws of physics" are suspect to scrutiny. Although in their case, they have already been scrutinized so much that we can all safely agree we have them down with 99.999% confidence. Still, the point is that with this method, it is impossible to prove things for sure. All that can be said is that they have not yet been disproved at present, but they certainly may be in the future.
If that's the case, then why should the scientific method be used anyway? Isn't it desirable to know things for sure? It depends on who you ask. I can think of many cases in which people probably don't want to know for sure. In science, not knowing for sure is what motivates its practitioners to continue making new discoveries and re-testing old ideas. Ironically, in religion, it is probably not desirable to know for sure. People believe God exists, but what if one day someone comes along and claims that she knows God exists? If the basis for religion moves from being a hypothesis to being a more or less guaranteed fact, what will happen to religion? Even though having faith means that one believes the tenets established by the religion to be true, I do not think those with faith can face the knowledge that it is true. If it is true, then where is the value or redemption found in believing?
Is it ever good to know for sure? I cannot answer that with certainty (haha, lame joke here), but I can certainly think of situations in which certainty is desired over proper methodology. The example I am thinking of goes back to the heading of my entry: Why an honest politician can never get elected. Democratic politics today is awash with politicians who seem to know just the thing that the nation needs to fix its problems. They take to the road and campaign with promises of programs and initiatives that will solve all problems. All the people have to do is to trust them and elect them into office. Since anyone can make unsubstantiated claims, the game goes to the side who can woo the voters with promises of a rosier future. The result of this condition is that once in office, the elected become faced with the tough reality of governance and soon realize that the promises cannot be kept. Is there a way to fix this problem? Let's think about this for a minute.
Imagine two politicians running for some office. One says, "I know our district is going through some tough times right now. Many people don't have jobs. Crime runs rampant through our streets. Our children are not prepared to go out into the real world because our school system is broken. Times are rough, but I have a vision for our district. If you elect me into such and such an office, I promise to put into action initiatives that will reduce taxes, encourage business and job growth, strengthen our police force with the funding they need to fight crime, and fix our schools so that our children can enjoy a better future. Here is my plan: (plan here.)" When asked how he plans to deal with the rising costs of healthcare, he answers immediately, "I will do this, this, and this." The people are impressed.
The next day, the other politician running for the same office comes into the same town and addresses the same audience. He gives his speech addressing the problems similar to his opponent. During the Q & A, someone asks the same question about healthcare. Pausing to think about it for a moment, he then answers, "Well, I know the rising cost of healthcare is a growing concern for workers and employers in our district, and it's a very complicated issue. Honestly, I don't have a definite plan to solve it yet, but I will say that I will look into the matter very carefully. I promise to collect the input of all the people in this area and consult the experts. After that, my aides and I will put forth what we feel is the best plan for our district and have an open debate about its merits and shortcomings. If the plan is fundamentally sound and receives broad based support by the people, our government will put it into action. Afterwards, we will come back and evaluate the plan's performance and determine whether or not modifications need to be made." I pose this question: Who do you think will get elected?
Stop and think. Should the solution be more sacred than the process? I don't think honest politicians will ever be elected in this country.
On a personal note, it has been an unusually warm winter here in Philadelphia. I had promised the HK exchange student that the winters here were bitterly cold. He must be disappointed, and so am I.
Friday, February 03, 2006
Harnessing the power at the edge of chaos (or how to embrace the 'E')
Today is the first Friday of the month, which means that the all important unemployment rate figures for January are released at 8:30am (yes, yes, I was there in front of a computer waiting for it to come out). The unemployment rate fell to 4.7% from 4.9% back in December. This figure came in under the market expectation of 4.9%, so one would be quick to think that the recent news would be a positive surprise for stock prices. If only it were that simple... Remember two things about stocks. First, their prices equal the present value of the future cash flows of the firm discounted at some nominal interest rate. Second, stock ownership represents ownership of a firm's real assets, whose real returns should not be affected by inflation in the long-run. The problem here is that in the short-run, inflation can and does hurt real values since it can distort the taxes faced by a firm on their profits.
Putting these two ideas together, it can be seen that there are two competing effects that determine the movement of stock prices. First, there is the dividend/cash flow effect. Increases in the working population under the market system suggest that firms are hiring more labor since the outlook for the future looks optimistic. Revenues and profits from operations are expected to increase, and so the shareholders that have the residual claim on those profits will find their shares more valuable. Second, there is the discount effect. Increases in the number of workers may also be inflationary. As the number of workers hired by a firm increases there is a decreasing marginal benefit to the extra labor hired. Looking at the changes in costs per unit labor as well as productivity growth sheds light the advisability as well as the sustainability of decreasing unemployment.
Yesterday, the quarter on quarter (qoq) productivity growth figures came out for Q4 of 2005. It was a dismal -0.6% (I believe I mentioned that they "really bit the dust" in yesterday's post.). Overall, the annual growth rate of productivity in 2005 was 2.7%. However, the qoq figures for unit labor costs rose 3.5 % with the annual rate coming through at 2.4%. This is bad news for profits. While the decrease in productivity was expected due to the low qoq GDP numbers released last week, this works against the dividend/cash flow effect. In combination with the unit labor cost growth (which by the way is the highest since the boom in 2000) that puts pressure on inflation, the decline in productivity may indicate a slowdown in the near future. Since movement in productivity growth (the first derivative) closely relates to GDP, it is considered a pro-cyclical indicator.
Now it's time for the main event: harnessing the power at the edge of chaos.
If I learned anything from classes in statistical inference, it is that
F(x) = B0 + B1*x + E
This here is the linear regression formula that is so often used and abused in statistical analysis. That little E at the end of the equation is a random error term that says "I know this regression line does not accurately (never at all) predict the observed values, so I am here to insert some randomness into the equation to make up for this." At its heart, all that regression wants to do is to eliminate that little E. That last term is only there to make things hard. "If I can only get rid of that E, won't it be great?" people running regressions think, but the fact is in the real world randomness is all around us, especially when the subject of scrutiny are human beings. May I suggest a slightly heretical thought here? Maybe, just maybe, we should think about things in a different way from what regression analysis tells us to do. Instead of trying to shove E out the door, why don't we throw out the betas and x's and embrace randomness? Maybe life is simply too complicated to be reduced down to a linear relationship.
Enter probabilistic modeling. The proposition is simple. Observations can lie, but perhaps it is possible to derive the underlying mechanism that is producing these observations with some careful thinking. At its heart, probabilitistic models tell stories of people acting on the individual level. They start by modeling the individual decision each person must make. From there, distributions arise that replicate the same process that produced the observations in the first place. The power of these models is that they can be modified to release the model-maker from a variety of unrealistic assumptions. For example, people are heterogeneous in their behavior. Probabilistic models do not require homogeneity. There is no "equal variances" assumption to be found here. We let the model and the data tell the story, instead of trying to find ways to reconcile the two.
There is another reason for the use of these models. One of the earliest lessons encountered in regression analysis is the inadvisability of making extrapolations. Trying the predict values beyond the range of the data already collected is a haphazard process that more often fails than not. Probability models do not face such limit restraints. Their predictive power far outweighs those of regression models. The reason is simple. What probability models estimate are the latent behavioral traits of the observed. As long as the underlying distributions of these latent traits do not undergo major changes, it is possible to make highly accurate prediction ad infinitum. In business, especially marketing, people are not concerned about how well your line fits or how high the R-square is. They care about things like forecasts, reach, and penetration. Probability models have the ability to deliver on many of these things far better than regression can ever hope to do.
Once again, the world is far from linear. Embrace the chaos, because God does play dice.
Putting these two ideas together, it can be seen that there are two competing effects that determine the movement of stock prices. First, there is the dividend/cash flow effect. Increases in the working population under the market system suggest that firms are hiring more labor since the outlook for the future looks optimistic. Revenues and profits from operations are expected to increase, and so the shareholders that have the residual claim on those profits will find their shares more valuable. Second, there is the discount effect. Increases in the number of workers may also be inflationary. As the number of workers hired by a firm increases there is a decreasing marginal benefit to the extra labor hired. Looking at the changes in costs per unit labor as well as productivity growth sheds light the advisability as well as the sustainability of decreasing unemployment.
Yesterday, the quarter on quarter (qoq) productivity growth figures came out for Q4 of 2005. It was a dismal -0.6% (I believe I mentioned that they "really bit the dust" in yesterday's post.). Overall, the annual growth rate of productivity in 2005 was 2.7%. However, the qoq figures for unit labor costs rose 3.5 % with the annual rate coming through at 2.4%. This is bad news for profits. While the decrease in productivity was expected due to the low qoq GDP numbers released last week, this works against the dividend/cash flow effect. In combination with the unit labor cost growth (which by the way is the highest since the boom in 2000) that puts pressure on inflation, the decline in productivity may indicate a slowdown in the near future. Since movement in productivity growth (the first derivative) closely relates to GDP, it is considered a pro-cyclical indicator.
Now it's time for the main event: harnessing the power at the edge of chaos.
If I learned anything from classes in statistical inference, it is that
F(x) = B0 + B1*x + E
This here is the linear regression formula that is so often used and abused in statistical analysis. That little E at the end of the equation is a random error term that says "I know this regression line does not accurately (never at all) predict the observed values, so I am here to insert some randomness into the equation to make up for this." At its heart, all that regression wants to do is to eliminate that little E. That last term is only there to make things hard. "If I can only get rid of that E, won't it be great?" people running regressions think, but the fact is in the real world randomness is all around us, especially when the subject of scrutiny are human beings. May I suggest a slightly heretical thought here? Maybe, just maybe, we should think about things in a different way from what regression analysis tells us to do. Instead of trying to shove E out the door, why don't we throw out the betas and x's and embrace randomness? Maybe life is simply too complicated to be reduced down to a linear relationship.
Enter probabilistic modeling. The proposition is simple. Observations can lie, but perhaps it is possible to derive the underlying mechanism that is producing these observations with some careful thinking. At its heart, probabilitistic models tell stories of people acting on the individual level. They start by modeling the individual decision each person must make. From there, distributions arise that replicate the same process that produced the observations in the first place. The power of these models is that they can be modified to release the model-maker from a variety of unrealistic assumptions. For example, people are heterogeneous in their behavior. Probabilistic models do not require homogeneity. There is no "equal variances" assumption to be found here. We let the model and the data tell the story, instead of trying to find ways to reconcile the two.
There is another reason for the use of these models. One of the earliest lessons encountered in regression analysis is the inadvisability of making extrapolations. Trying the predict values beyond the range of the data already collected is a haphazard process that more often fails than not. Probability models do not face such limit restraints. Their predictive power far outweighs those of regression models. The reason is simple. What probability models estimate are the latent behavioral traits of the observed. As long as the underlying distributions of these latent traits do not undergo major changes, it is possible to make highly accurate prediction ad infinitum. In business, especially marketing, people are not concerned about how well your line fits or how high the R-square is. They care about things like forecasts, reach, and penetration. Probability models have the ability to deliver on many of these things far better than regression can ever hope to do.
Once again, the world is far from linear. Embrace the chaos, because God does play dice.
Thursday, February 02, 2006
Physics to the rescue (Many thanks to Feynman)
My fate was sealed well before the close today. What a spectacular failure it was all along the way! Losing 1.6% isn't amusing (it's even harder to stomach when you look at the absolute values). I smell some bargain hunting tomorrow when the market opens again; however, things may not be so simple depending on the employment numbers that come out on Friday.
At the macroeconomic level, things aren't looking so good going forward. The labor productivity growth numbers for the last quarter really bit the dust when they came out today. Preliminary measures indicate that it's actually negative qoq, which suggests that employers are hiring workers with a marginal benefit that's decreasing faster than the market anticipates. Due the forward looking nature of stocks, their values will adjust accordingly to reflect the new outlook, especially if unemployment falls. That would be a bad combination. Grim, grim stuff.
Now, let's get to the main topic for today. Sitting in FNCE 101 today, I found my thoughts meandering from the beaten path Dr. Souleles was meticulously laying out for me. It was on this road less traveled that I had my inspiration for the day, so here I document it for the record.
I recalled a conversation I had with my language teacher two days back. We were discussing the notions of rulership passed down since antiquity in China. She laid out the case that the mindset of Chinese officials is to administer the state as if it was a large family. While this family is diverse ethnically and geographically, the concept of state and household is intimately related - just ask a Mandarin speaker how to say the word for state or country. The paternalism of the Chinese system is simply a natural manifestation of this concept of rulership. There's nothing wrong with the way things are when one takes this into consideration.
At the time, I quietly listened to what she had to say and pondered these things in my heart. There was something that unsettled me about the bland acceptance of this idea, but I could not clearly identify what it was. How funny was it then I came up with a rebuttal by drawing from concepts found in physical law (while listening to a lecture on finance no less)! I distinctly remember in my brief foray into the world of mechanics the concept of symmetry. Symmetry as defined in physics is basically invariance under any kind of transformation. There are many types of symmetries to be found in physics: translational, rotational, temporal. A transformation is said to exhibit symmetry when an action or experiment can be carried out with the exact same results before and after the transformation, providing that all the relevant physical forces and conditions are the same before and afterwards.
Why do I bring up symmetry? Because there is a type of transformation which does not exhibit this property. Transformations of scale do not. An experiment performed at the microscopic level is not guaranteed to produce the same results at the macroscopic level, even if all the relevant forces and conditions have been perfectly enlarged to the same degree. The reason behind this is actually quite simple. The number of atoms is different before and after the transformation. Increasing in size also means taking on more atoms, since atoms don't grow with the transformation in scale. This results in different dynamics at differing levels of magnitude.
Would it be a far stretch to say that not only physical law acts this way, but also social order? Even the large extended families that existed in China's past do not come close to being close to the size of a nation. One may liken them unto atoms of this great something called China. While there may be social norms and rules that direct the daily lives of the individual family members and give authority to some over others, to say that one can and should govern the state as if it was a large family would be erroneous. In a society with hundreds of millions of families, the entire dynamics of their interaction changes. The concept of benevolent parenthood cannot be stretched this far as to encompass proper rulership. At some point, it will break down. You cannot settle everything as a family affair because at some point people cease to interact in such a way that they consider each other intimately related. This is why the rule of law is so critical to a well functioning society. Let me just say: Tianxia fei yi jia. Zhijia fei zhiguo, zhiguo fei zhijia.
At the macroeconomic level, things aren't looking so good going forward. The labor productivity growth numbers for the last quarter really bit the dust when they came out today. Preliminary measures indicate that it's actually negative qoq, which suggests that employers are hiring workers with a marginal benefit that's decreasing faster than the market anticipates. Due the forward looking nature of stocks, their values will adjust accordingly to reflect the new outlook, especially if unemployment falls. That would be a bad combination. Grim, grim stuff.
Now, let's get to the main topic for today. Sitting in FNCE 101 today, I found my thoughts meandering from the beaten path Dr. Souleles was meticulously laying out for me. It was on this road less traveled that I had my inspiration for the day, so here I document it for the record.
I recalled a conversation I had with my language teacher two days back. We were discussing the notions of rulership passed down since antiquity in China. She laid out the case that the mindset of Chinese officials is to administer the state as if it was a large family. While this family is diverse ethnically and geographically, the concept of state and household is intimately related - just ask a Mandarin speaker how to say the word for state or country. The paternalism of the Chinese system is simply a natural manifestation of this concept of rulership. There's nothing wrong with the way things are when one takes this into consideration.
At the time, I quietly listened to what she had to say and pondered these things in my heart. There was something that unsettled me about the bland acceptance of this idea, but I could not clearly identify what it was. How funny was it then I came up with a rebuttal by drawing from concepts found in physical law (while listening to a lecture on finance no less)! I distinctly remember in my brief foray into the world of mechanics the concept of symmetry. Symmetry as defined in physics is basically invariance under any kind of transformation. There are many types of symmetries to be found in physics: translational, rotational, temporal. A transformation is said to exhibit symmetry when an action or experiment can be carried out with the exact same results before and after the transformation, providing that all the relevant physical forces and conditions are the same before and afterwards.
Why do I bring up symmetry? Because there is a type of transformation which does not exhibit this property. Transformations of scale do not. An experiment performed at the microscopic level is not guaranteed to produce the same results at the macroscopic level, even if all the relevant forces and conditions have been perfectly enlarged to the same degree. The reason behind this is actually quite simple. The number of atoms is different before and after the transformation. Increasing in size also means taking on more atoms, since atoms don't grow with the transformation in scale. This results in different dynamics at differing levels of magnitude.
Would it be a far stretch to say that not only physical law acts this way, but also social order? Even the large extended families that existed in China's past do not come close to being close to the size of a nation. One may liken them unto atoms of this great something called China. While there may be social norms and rules that direct the daily lives of the individual family members and give authority to some over others, to say that one can and should govern the state as if it was a large family would be erroneous. In a society with hundreds of millions of families, the entire dynamics of their interaction changes. The concept of benevolent parenthood cannot be stretched this far as to encompass proper rulership. At some point, it will break down. You cannot settle everything as a family affair because at some point people cease to interact in such a way that they consider each other intimately related. This is why the rule of law is so critical to a well functioning society. Let me just say: Tianxia fei yi jia. Zhijia fei zhiguo, zhiguo fei zhijia.
Wednesday, February 01, 2006
This is just a test... Right
After spending nearly a thousand days on xanga, I've finally found the resolution to start posting from my personal website. More for myself than others, this blog is for meticulously documenting my mental and emotional state as well as other trivial and transient things like how my stock did today or what I found left behind in Huntsman Hall. Stick around for a while and (time permitting) browse the other pages which also contain information that I have foolishly placed in the public domain.
I will enjoy one day looking back at the foolish things I've written and the even more ridiculous ideas and beliefs I entertained. I hope readers would too.
I will enjoy one day looking back at the foolish things I've written and the even more ridiculous ideas and beliefs I entertained. I hope readers would too.
0 Comments:
Post a Comment