Utilizing Reliable Data for Effective Decision Making

Finding and Using Reliable Data 1 April 14, 2023 Craig R. Parsons, YNU

Quantitative and Qualitative Data 2 Today s talk will be about Quantitative data (GDP, Prices, Trade figures) But your paper might use qualitative data too. Craig R. Parsons, YNU

Lies, Damned Lies and Statistics (PM Disraeli) 3 Data can be very important to validate or refute theory. But: Have a healthy skepticism of any data; and Remember data may be consistent with more than one theory. Just because your data seems to match with some theory does not GUARANTEE that theory is correct! E.g. maybe you find a strong correlation between drinking red wine and lower heart attacks. So, Theory 1: drinking wine lowers chance of heart attack might be correct. But Theory 2: might be too: People who drink red wine tend to have higher income than those who do not, and those with higher incomes can afford better health care and so have less heart attacks. (Omitted variable bias, and/or sample selection bias.) Both would show positive correlation between drinking wine and lower heart attacks: If you believe Theory 1, you might (incorrectly) encourage people to go an drink more red wine! (rather than going to the doctors more often for check-ups, etc.) Craig R. Parsons, YNU

Skepticism of Data 4 The government are very keen on amassing statistics. They collect them, add them, raise them to the nth power, take the cube root and prepare wonderful diagrams. But you must never forget that every one of these figures comes in the first instance from the village watchman, who just puts down what he damn pleases. Josiah Stamp Craig R. Parsons, YNU

Best data available 5 No data is perfect. But data by WB, IMF (and US govt) are generally the best you will get. They have teams of professional statisticians gathering data in the best ways possible. But again. No data is perfect. Other countries data, of course, can often be quite good. To the extent possible consult experts in that field to assess data reliability. (Alternatively, read lots of paper in that field to see what data the top researchers rely on most.) Craig R. Parsons, YNU

Use the same source for each series 6 Don t splice data series. E.g. you have data for car imports from 1995-2010 from UN Comtrade but then the data does not go back further than 1995. But, elsewhere (maybe US data), the same car import data exists from 1980-1998. Don t mix or splice this data. (Unless you have very convincing overlaps which matches exactly. For different series, using different sources are fine. Craig R. Parsons, YNU

Real vs. Nominal 7 If you have time series data (GDP, FDI, etc.) make the series are all either nominal, or all real (depending of what you are doing). Don t MIX nominal and real series. (e.q. real GDP, but nominal FDI.) Remember: distance, number of people, etc. are real variables. Also, when using data which is quarterly or monthly be aware if the data is or not seasonally adjusted. Don t mix adjusted and unadjusted series. Craig R. Parsons, YNU

On Input Output Data 8 Please consult Profs. Uemura, Shrestha, Ujikawa, or Ishiro. Two major sources of I/O data are: WIOD and our (YNU s GIO) international database. WIOD is here: http://www.wiod.org/new_site/data.htm YNU s GIO is here: http://www.recessa.ynu.ac.jp/en/modules/ynugio/index.php?con tent_id=1 Craig R. Parsons, YNU

For world trade data 9 UN COMTRADE, UN TRAINS (tariff data too) and UNCTAD are very good. (Penn World tables, Barro and Lee also good; the latter for educational data.) For individual countries, often the trade bureau has more detailed (9 digit) level. US, Japan and EU (Eurostat) are have high quality data. Other countries do too. (Trick: if you don t trust the Vietnamese export data to Japan, then use the Japanese import data FROM Vietnam, gathered by Japan Customs.) Craig R. Parsons, YNU

On Foreign Direct Investment (FDI) data 10 Beware of using FDI flow or stock data (from UNCATAD or anywhere else) rather than activity by affiliates for a measure of FDI activity. Financial FDI flows data is exactly that, financial flows. Often not a very good indicator or real economic activity by MNCs. Unfortunately activity data is much harder to come by. Craig R. Parsons, YNU

Gathering your own data 11 If you conduct your own survey, great! This is a great way to ADD knowledge to the world. If you do, make sure you try to construct the survey in careful (clever) way. Pre-surveys attaining HIGH response rate. Being aware of sample selection bias E.e. if you are interested in seeing how many people evade their taxes, maybe only the honest people answer your survey! Bias! Craig R. Parsons, YNU

Financial/Stock price data 12 For those interested in stock prices, daily returns, etc. some of the is data is available online for free. But sometimes it is not. It may be available, by some data collection company, for a very high price. If so, YNU may have it. But probably not. You might want to change your topic a bit Or dig in newspapers. Daily stock prices DO exist it is just that people nowadays are too lazy to do digging and copying from old newspapers. On that theme: much data IS out there but not yet digital. This is here YOUR sweat can produce new data. (Especially firm data.) Craig R. Parsons, YNU

Look where the light is 13 Don t torture the data and try to force it tell you something it really doesn t know. Don t try to estimate a money demand function for a country which only has 10 years or annual data and three civil wars over that time. Don t try to replicate a study on FDI activity by Japanese firms for, say, Sri Lanka firms, if you don t have the data. (If you can do a survey and MAKE the data, great!) To put it another way: a lot more data exists for high-income countries than for low-income countries. This is an unfortunate reality, which hopefully is changing. Craig R. Parsons, YNU

Dont try to answer BIG questions for your thesis. Little questions are fine. 14 E.g. Don t try to estimate an equation a what makes my country grow faster? Try to be more specific and narrow, where the data exists. Craig R. Parsons, YNU