Mostrando entradas con la etiqueta tips. Mostrar todas las entradas
Mostrando entradas con la etiqueta tips. Mostrar todas las entradas

sábado, 16 de agosto de 2014

Leyendas y dragones de la Econometría

Magos, espadas y rosas: leyes y leyendas de la econometría
por Walter Sosa Escudero


Una vez alguien puso en un blog de matemática la siguiente pregunta ¿Hay algún número que es tan pequeño que uno pueda tomarlo como si fuese cero? A todos nos suena a disparate esta pregunta, pero la vida cotidiana está llena de reglas y atajos que nos permiten proceder con comodidad. A modo de ejemplo, si alguien me pregunta mi edad, respondo “48 años”, y no “48 años, 8 meses, 12 dias, 6 horas, 26 minutos, etc.”, porque hemos convenido, sin firmar papel alguno, que la edad se pregunta y responde en años. Entonces, cualquier regla es un intento de resumir algo complejo, lo cual es necesariamente bueno y malo, tan malo como que mi edad no es exactamente 48 años, y bueno como que la discrepancia entre mi verdadera edad y 48 es socialmente insignificante, y es lo que permite comunicarnos sin volvernos locos. Ahora ¿Dónde está escrito que la edad se comunica en años? ¿Y las distancias en cuadras?
La econometría no está exenta a esta dinámica. Existen varias reglas, que mucha gente usa pero que ningún teórico de la econometría admite haber pronunciado. Entonces, vaya aquí una colección incompleta y disparatada de “reglas econométricas”, que todos usan y nadie admite. Deliberadamente omitiré las fuentes, por por lo menos una de las siguientes razones (no mutuamente excluyentes): 1) Se dice el pecado, pero no el pecador, 2) Las desconozco, 3) Las conozco pero los que las inventaron son prestigiosos colegas a quienes no quiero desprestigiar. Solo daré algunas pistas para que en estos tiempos berreta de Google, hagan su propio trabajo arqueológico.
1. n>30. Quizas la más vieja de las reglas econométricas. Muchos estudiantes usan esta regla para hablar de “muestra grande”, es decir, en esta lógica maniquea 29 es “chico” y 31 es “grande”. Cuando en mi curso básico pregunto “cuán grande debe ser una muestra para ser considerada grande” (una pregunta bastante pavota, por cierto), por alguna razón esotérica mucha gente responde “más de treinta”. El autor de la regla ha sido injustamente olvidado. O quizás todo lo contrario.
2. T>20. Este es el límite de observaciones temporales más allá del cual el sesgo por panel dinámico no debería molestar. Es decir, con T>20 uno podría usar el estimador de efectos fijos y dejarse se embromar con Arellano-Bond y otras cosas estrambóticas. Esta cifra fue recientemente actualizada (a 30) en un paper de Judson y Owen. Los autores de la regla “abandonaron la institución, nunca más se supo de ellos, del caso no se habló más” como decían los Fabulosos Cadillacs de los oficiales que vieron morir a Manuel Santillán, El León.
3. G>50. Se dice por ahí, en los bajofondos de la econometría, que con más de 50 clusters los estimadores “cluster robust” son confiables y que con menos hay que apelar correcciones por sesgos, wild bootstrap y otras invocaciones mefistofélicas. Con mas de 50, con vce(cluster) alcanza. Ahora, ¿no es realmente sospechoso que 50 se parezca, misteriosamente, a la cantidad de estados que hay en los Estados Unidos de America? Mmm.
4. (p=f+1). Esta regla dice que la cantidad de rezagos (p) a incluir en un modelo dinámico (o en un ARMA) es la frecuencia de los datos (f) más uno. Por ejemplo, con datos trimestrales, f=4 (cuatro trimestres), ergo hay que incluir 5 rezagos. Y la cuenta da 13 para datos mensuales (y 2 para anuales). Esta regla me fue comunicada en una intersección de dos caminos de tierra que no figuran en ningún mapa, en el estado de Illinois, en una media noche de luna llena. Había un tipo con una guitarra, me escondi hasta que se fuera y recién ahí me aposté con mi notebook y mi PCGive.
5. F>10. Esta está en el libro de Angrist y Pischke, pero ellos no dicen quien la invento (asi cualquiera, que vivos!). Le regla reza que si el F de la primera etapa es mayor que 10, no hay un problema de instrumentos débiles. Ahora, si F es menor que 10, terribles cosas pueden suceder. Me contaron de un econometrista de Springfield, Arkansas, que quiso publicar un paper con F=9.99, tras lo cual su esposa lo abandonó, perdió toda su fortuna y ahora se dedica a escribir cadenas de mensajes por Facebook.
6. l= 1600. Dice la leyenda que Prescott se negaba sistemáticamente a publicar el famoso paper del “filtro de Hodrick y Prescott” (escrito en 1981 y publicado recién en…1997!) porque lo consideraba demasiado trivial. Como todos saben, el filtro necesita la previa determinación de un parámetro de suavizado, que los autores sugieren fijarlo en 1600, para datos trimestrales. Se cuenta de un grupo de econometristas dinamarqueses que fijaron l=1613, y que en el intervalo de un congreso en donde iban a presentar sus resultados, adujeron salir por unas cervezas, tras lo cual nunca más nadie oyó hablar de ellos.
Como las brujas, las reglas no existen, pero que las hay, las hay. ¿De cuales me olvide?
Buenas noches.

Econometría Avanzada

jueves, 26 de septiembre de 2013

15 consejos poco éticos para zafar algunas veces

15 More Unethical And Illegal Ways That People Get Ahead In Life



Based on the reaction to a previous post it's clear that people are very interested in the dubious and unethical ways that people save money and cheat the system to get ahead.
We've pulled together a new thread of dirty tricks and illegal cheats from Reddit and elsewhere.
These hacks are uniformly unethical. Some veer into fraud. They should be regarded as informational and for entertainment purposes rather than as actual suggestions. You shouldn't do them. If you do do them, you are a bad person.
"At my summer theater where everyone is brought in from various parts of the nation and lives together for the season, the length of employment is roughly 68 days. Walmart's electronic return policy is 90 days. So, the carpenters would get a huge widescreen hi-def, whatever, lovingly store every scrap of packaging and the receipt, and take it back as soon as the season was over. It was actually pretty freaking smart."
walmart
Walmart Media Relations
Via Reddit:
"If you want to cancel your cellphone contract without paying a fee, pull up the provider's service map. Find a huge hole in the map, like a desert out west. Look for a town name in that map. Tell them you're moving to Putzachateeawaka, Arizona and you want to cancel because they don't provide service there. Boom.
"3 times now. 3 times."
Ruta 40 Argentina
Shutterstock
"Putzachateeawaka" is not a real place, but this is a remote desert that probably doesn't have good cell coverage.
Via Reddit: 
"This one should be kept a secret but whatever. Next time you're at your dentist or any place that has a layout of various magazines, take a picture of the barcode, address, and name. When you get home that's all the information you'll need log into the online version of the magazine. Either that or set it up on your iPad and enjoy free versions of magazines forever." 
Building a startup or website? There are lots of underhanded ways to boost traffic, via OMGPOP founder Dan Porter:
  • Drive traffic, either directly or indirectly, with pornography. "It's cheap to buy and can generate a lot of clicks," Porter writes.
  • Traffic robots and meta-refresh can help artificially boost stats as well. 
  • More suggestions here.
Via Reddit user eightbitrob:
"I've been canceling and switching the name on my cable bill now for 10 years between myself and my wife so I always get the new customers rate."
Via Reddit user slainte 99:
"I know a married couple who started a "food blog" basically so they can write off their entire grocery bill."
It's surprisingly easy to take advantage of people's fear and insecurity with the "fear then relief" technique. Basically, you make someone think they're in serious danger of losing their livelihood, or anything else likely to provoke fear. Then you relent, and ask for something less. They're much, much more likely to comply
An example would be a manager making an employee think that their job's in danger, leaving them to stew for a few hours or a day, then relenting and letting them know they're safe. That's when they ask for them to work overtime. 
undercover boss boston market
Boston Market
Via Reddit user zippytuna:
"I learned that you can get into almost any special event by wearing a chef coat. Even just carrying one and walking like you know where you're going will work every time. Most people don't want to look stupid by asking you who you are." 
An unethical business tactic that isn't technically illegal, via Quora user Antone Johnson:
"(Tie) up a key competitor or strategic partner in extensive negotiations (merger, joint venture, etc.) when you aren't seriously considering actually doing the deal and are just trying to learn as much as possible about their business -- or use them as a competing bidder to drive up the price for the other company you're serious about."
Via Reddit user aawv:
"My go to missing work call was never "I'm sick", it was "Family problems". They never questioned it, it's vague enough and embarrassing enough that nobody ever asks."   
Lots of people have figured out how to look busy and satisfy their bosses without doing all that much work. A few tips to accomplish this.
  • Respond extremely rapidly to emails. It doesn't take that much effort, and makes you look on top of things.
  • Prioritize visible tasks for the people that can help your career over anything else, regardless of import or job function.
  • Always have a long list of things you're working on so you aren't assigned more.
  • Pad your estimates of how long things will take by 25%, and deliver ahead of schedule, but use this sparingly.
  • Make a "decoy screen."

working, busy, focused, tumblr, january 2012, bi, dng
Daniel Goodman / Business Insider
Via Reddit user DylanAnderson:
"My friend from high school works at the recycling center for his college. About a thousand books get thrown away each semester because the school won't buy them back. Amazon definitely buys them though. He gets about a $1000 bonus every semester."
Via Reddit user irontumbleweed:
"When career hunting, flagging Craigslist posts as spam after applying, or taking down flyers for a position (like at a college campus). cuts down on the competition."
"Lots of government bodies/agencies buy computers with the OS pre-bundled and pre-installed (Windows Vista, 7, 8) which they then overwrite with XP because that's what the IT department wants. If you see a relatively new computer running XP, check the side of it for the license sticker. More likely than not it will be a perfectly valid windows 7/8 pro license key not being used and never going to be used. I've picked up over 50 this way. Anytime I need to upgrade a friend's PC, I use one. No point in them going to waste."
Via Reddit user chrisw7128:
"I always sleep in the Islamic prayer rooms when I have a red-eye layover somewhere...they're fairly spacious, absolutely quiet, and clean as can be. There is plenty of room to lay down on the carpet, and security never comes in to bother you."

miércoles, 24 de julio de 2013

Tips para pasantes en Economia

Internships in Economics – Lessons from an Economist!

written by Samridhi Shukla


Whether pursuing a bachelor’s degree in economics or having finished a PhD from a top university, at all stages of the academic pursuits and research careers, young economists have been engaged in the grueling process of finding internships with reputable organizations. While masters and bachelors students aim to gather work experience and transition into full time jobs, PhD candidates attempt to combine their research with the work of relevant organizations.
Having gone through the cumbersome process myself and having successfully crossed over into a “real job” after a tedious string of internships, here are a few lessons and tips I would like to share with you, no matter which part of the internship journey you are currently navigating through.
Lesson 1: Do what interests you!
It is important to go for an internship in a discipline that really interests you. In the long run companies will hire you for the skills you can offer them and this decision will not be based on the reputation of the company where you did your summer internship.  Work experience in a local unknown NGO in Congo is as valuable as an internship at the Big 4 consulting firms in New York. Bottom line, do something you are good at, pick out solid skills you can apply in life and don´t make a company´s reputation the top factor in your decision.
Lesson 2: Building expertise or versatility?
You can stick to the same area of interest as you proceed or you can be versatile in your choice of internships! There is no golden rule about what works.  However, anecdotal evidence suggests that sticking to one area of interest builds a certain credibility of expertise in that area and may lead to brownie points for you from recruiters who want specialists. Having said that, I firmly believe that while in university, one should have full freedom to taste different worlds and thereby realize with experience what they would ultimately like to do. As long as you can explain your path and give convincing reasons for the choices you made, no experience is inferior to another.
Lesson 3: The money will come as you get better!
I remember vividly how unpaid internships were looked down upon during my university days. People make you believe that you are selling yourself too low if a company where you intern does not offer you remuneration. Please do not believe this especially if you are at the start of your internship phase in university! As a natural progression, I can guarantee you that the money will come as you get better at what you do, i.e. chase excellence and not money and sooner or later, money will chase you for your excellence. In fact, sometimes the companies that offer you the best training and work experience do not have the budget to support interns. However, with what they teach you, your skills might be worth a lot once you do enter the real job market.
Lesson 4: Most important: Do not compare!
I am going to repeat here what granny said all along. Do not compare yourself with others! Though it sounds cliché, this is crucial.  While I was in love with Development Economics and not particularly fond of the mathematical methods of economics, I made myself go through several unnecessary and painful processes of applying for finance based positions in leading investment banks all over the world. Why? Peer-pressure in a top economics college of India where everyone blindly runs behind big brand names offering high salary packages! Of course, the outcome was repeated rejection- not good for one´s self esteem and also not worth it in the long term, as I can see today in retrospect.
In conclusion, I would like to say that what you make out of an internship experience and how much you leverage from the social network you make depends on you. Every experience contributes to who we are as people and professionals. Make sure you take time to dig into the details of what you want, what you learned from a given experience and share it with the recruiters in an authentic way as you go along. At the end of the day, recruiters want authentic people with authentic experiences and real life skills. Gather those and move ahead with confidence!
Good Luck.
 Image credit:  taniawillis