Exploring our resources - try the Business and Management Resources page (LibGuides), scroll down to category or tag cloud in the right-hand column, search this blog (top right) or search Manchester Business Answers 24/7 (FAQ). For latest news see @MBSLibrary Follow @MBSLibrary
In 2013 we had 64 new posts, just over 110,000 views, and All UK listed companies (part 1) was the most commented post. Of the top 5 posts Journal Ranking – marketing was from 2012, three were from 2011, and one from 2010. Please read Reflecting on Business Research Plus and comment with your suggestions for improvement. [22 January 2014]
There are many ways to bring your data into IBM SPSS Statistics, for whatever manner of analysis or reorganisation you wish to perform. Here are a few approaches to consider, with some of their relative merits and shortcomings. Please read Part 1 of this blog post before proceeding, which covers opening an existing data source in SPSS Statistics format and typing or pasting in data.
Approach 2: Read text data directly from other formats
You can open data files created in other applications, from the File -> Open menu or the File -> Read Text Data menu. (If you spot any difference between these two, please comment below.) The file type choices include:
- Text (*.txt, *.dat, *.csv) — tab, comma or otherwise-delimited
- Excel (*.xls, *.xlsx, *.xlsm) — one sheet only
- SAS (*.sas7bdat or others) — one data set
- Stata (*.dta)
The plain text or Excel document may have variable names in the first row, but only the first row. It may instead have no variable names at all. It is worth preparing your source data file and checking that it has no surprises for SPSS like a variable type changing from 2 decimal places to no decimal places to text within the same column.
Preparing an Excel document
- The numerical data has 9 decimal places but is formatted to show none.
- The variable names are split over two rows and contain space characters.
- The data does not start in the top-left corner.
- The variable names are in the first row only, with no spaces or punctuation (although underscores and dots would be okay).
- The data starts from the first column (which contains an identifier “year”), and from the second row.
Example: reading text directly from Excel (quick)
In, SPSS, choose the File -> Read Text Data menu and select your prepared Excel document. You can choose if the file has variable names in the first row or now, choose which sheet to read from, and optionally specify a cell range and maximum column character width for string (text) columns.
After clicking on OK, the data will immediately appear in SPSS (see right). The variable types were guessed by SPSS based on the content of the first row of actual data, and the number of decimal places is set to 1 for numerical data. Although you can change the number of decimal places in the Variable View tab, the data past the first place is lost. The identifier variable “year” has had one decimal place added too, which should be removed too.
This method does not allow you to change the variable type or properties, so consider the longer approach outlined in Part 3, going via a plain text file.
This post will continue soon with Part 3.
IBM SPSS Statistics version 20 was used in this blog post, but the methods should apply to older and newer versions too. The data used in these examples comes from UK’s carbon footprint, 1997-2011, retrieved 3 April 2014.
A couple of queries recently have been about where to get “CEO compensation” data, and this turns out to be a good example where the answer depends on refining the initial question.
At the University of Manchester we have two specialist financial databases that cover executive pay, and therefore CEO (Chief Executive Officer) compensation.
- BoardEx - this is good for searching current board memberships, but one query was looking for 10 years of historical CEO compensation data for a list of companies
- Execucomp (Compustat Execucomp) – this is available on WRDS, would cope with historical data and a list of companies, but only if they are US companies.
Next I re-read the May 2012 post Executive Pay – this gives Thomson One Banker and Bloomberg as options.
Thomson ONE.com is similar to Thomson One Banker – you need to look up the executives, find the CEO on the list and then select to get the detailed information. (It is even harder if you want a previous CEO)
Bloomberg has a function MGMT that allows also allows you to browse company executives. However, it also has two variables for the total salary and the total compensation of the company CEO (or the executive Bloomberg judge as equivalent to the CEO). I used these in the Bloomberg Excel add-in after loading list of companies.
A little more skill with the Bloomberg Excel add-in and these results could have been presented in a neater fashion.
Datastream is usually good if you want historical numerical data, but searching using “compensation”, “pay”, “salary”, “CEO” all yield no evidence of a suitable time-series datatype.
So as I am looking for historical CEO compensation, including non-US companies, and can get to the library Bloomberg looks to be the best option.
A recent comment asked about creating a list on UK companies listed between 2000 and 2013 and with a market capitalisation of more that $10 million using Datastream. The basic technique is to use a Datastream static request to get data about UK companies, and filter to create the required list.
There are a number of key Datastream datatypes that commonly used
- DSCD – Datastream code
- EXDSCD – Exchange Datastream code (as a check)
- BDATE – Base date
- TIME – latest/final data of price data
- MV – market value (market cap), MV~U$ will give this is US dollars.
If you want companies listed between 2000 and 2013, is this companies that are listed continually from 01 Jan 2000 to 31 Dec 2013, or as is more common companies that are listed for some time between 01 Jan 2000 and 31 Dec 2013.
If you are looking for companies with a market capitalisation of more than say £100 million, is that companies whose market cap is always over £100 million, or those whose market cap is over £100 million at some time. The spreadsheet below illustrates one approach.
The spreadsheet above shows the results of a static request for BDATE and TIME and the market value in US dollars (MV~U$) on the 31 December for each year 2000 to 2013.
For the market value, an alternative is to make a time series request for the market value in US dollars, for example for each quarter 2000 to 2013, and combine this with the static request.
If we take the maximum of all the MV~U$ values in the spreadsheet above then we have the maximum market value at each year end. The maximum of the MV~U$ values for the quarterly time series gives the maximum at each quarter end for the relevant time period.
The Datastream predefined list WSCOPEUK is a reasonable choice for an initial list. (See earlier post All UK listed companies (part 2) for details) We can use Excel to sort and filter to give the required set of companies.
- 5189 – WSCOPEUK – UK companies
- 4535 of these – have a market cap in US dollars over 10 for at least 1 quarter end 2000 – 2013
- 3848 of these – have TIME > 01 Jan 2000, so were listed for at least part of the relevant time period
- These 3848 do include 8 companies that were only listed in December 2013 (BDATE > 30/11/2013) and 12 companies that were delisted in January 2000 (TIME < 01/02/2000)
Happy with this list of 3848 UK companies, then select the column of Datastream codes (DSCD) and use the Create List(from Range) option in the Datastream tab to create your list that can be used in future Datastream queries.
Betas represent the volatility of a company with respect to a market index.
Beta. The sensitivity of the share to market moves. A share with a beta of 1.0 tends to perform in line with the index; one with a beta of 1.2 tends to change by 1.2 percent for each 1 percent move in the index. (London Business School, 2012)
Beta values are used in many financial market models and therefore researchers often ask about which databases include betas. Thomson ONE.com has a relatively easy way of getting historical betas (and current betas if yesterday is current enough).
- Goto Thomson ONE.com - remembering to use IE (Internet explorer) – select the Company Views tab
- Lookup your company in the usual way and select Price Chart
- Select Beta Coefficient in one of the LOWER PANE SERIES drop down boxes (screenshot below)
You can use the View Data button to see the data values that have been charted, or the Excel icon to export the results.
Other sources of beta values include Bloomberg, Datastream, and LSPD for UK companies. There is a Betas Guide on the business and management guides page.
The basic formula for calculating a beta value is fixed but the details can vary so that different sources may give slightly different values. For example, for a UK company Bloomberg uses the FTSE 100 as the default market index and weekly data while Datastream uses the FTSE All Share index and monthly data.
London Business School (2012) “Guide to Company Tables”, Risk Measurement Service, 34(7), p. 9.
There are many ways to bring your data into IBM SPSS Statistics, for whatever manner of analysis or reorganisation you wish to perform. Here are a few approaches to consider, with some of their relative merits and shortcomings. This post is split over three parts.
Before you proceed, you should be at least slightly familiar with the main window in SPSS Statistics, the Data Editor. Specifically, there are two views as identified by the orange tab at the bottom-left of the screen: Data View and Variable View. The former has the variables in columns with observations/readings in rows; the latter has the variables in rows with their meta-data in columns.
Best-case scenario: Open an existing data source in SPSS Statistics format
You might be fortunate enough to already have data in the native format to SPSS Statistics (*.sav). This is the format to choose when saving your data while working in SPSS.
Each variable has a Name, and that name cannot contain spaces, punctuation (except dots or underscores) and cannot begin with a number. In older versions of SPSS, variable names could only be 8 characters long. It is good practice to use more explanatory Labels with your variables as well as short-hand Names. This will help you if you come back to your data in the future and cannot remember it as well as you thought (it happens to everyone!) or if you pass on your data to somebody else.
Approach 1: Type in data, or copy and paste
You may create a new, blank document and save it in SPSS Statistics default format. Set up your variables carefully, including the variable type (e.g. Number) and number of decimal places, before you type any data in. This is especially important if you choose to copy and paste your data in from another source such as a spreadsheet, or you risk your data being rounded down to integers.
Remember to use a numerical variable type wherever possible, even if your data appears to be in labelled categories such as Yes/No or UK/Europe/World. SPSS works best with numbers, so record your categories as integers (e.g. 0/1 for No/Yes) then assign value labels once it is in SPSS (that’s the Values column in the Variable View).
To toggle the display of the data category labels and the numbers behind them, go to View -> Value Labels when in the Data View tab.
IBM SPSS Statistics version 20 was used in this blog post, but the methods should apply to older and newer versions too. The data in the screenshots come from a British Crime Survey, 2010, and were prepared by The Cathy Marsh Centre for Census and Survey Research.
Just added the LSE Library Blog to the blogroll widget on the right side.
The LSE (London School of Economics and Political Science) Library Blog does a regular feature that highlights online resources of interest. The most recent Social Science sites of the week from LSE library (28 March 2014) includes sections on women on boards, engaging with massive online courses, research ethics guide book, and the Australian Government Web Archive (AGWA).
The blog also has an excellent post about the recent UK budget with lots of links to additional online resources – UK Budget: get the facts with LSE Library’s recommended academic resources (19 March 2014)
In addition, you have to like a blog that includes a post on the economics of chocolate - Celebrate Valentines day with our economics of chocolate special (14 February 2014)
If you are interested in the UK budget, there is a “TweetMap” and commentary from Anita Greenhill of Manchester Business School - The social media reaction to #Budget2014
For more on women on boards try the Corporate Law and Governance blog board diversity posts.
<HELP> X 2 – press the help key twice – to start a live instant message (IM) session with the Bloomberg helpdesk. This has been such a core part of the help available on Bloomberg Professional that it is a surprise to see it change.
<HELP> X 2 – now gives a screen that describes various ways of exploring Bloomberg’s online help and, at the foot, a link to “submit a question to the Bloomberg Help Desk and receive a response in one business day”
A number of academic Bloomberg subscribers have reported this change. However, there is no mention of it as a recent update so may not apply to all Bloomberg users. (Further evidence for this is that there are frequent references to the live help on Bloomberg.)
Following the Bloomberg Help Desk link you get a form that is very similar to the old IM form for live help. The differences include the email address field and the clear statement that the email response will be “by the next business day”.
When you submit you get a reference number.
The email query facility works. I submitted a question around 10:00am and got a response about 3 hours later.
The email reply does contain a paragraph that suggests this new procedure for contacting the Bloomberg help desk has not been thought through.
This response from the Bloomberg Help Desk has been forwarded to your external email address at your request. Please do not respond to this email. For a full log of this and other inquiries please login to the Bloomberg and run HDSK<GO> or for further assistance press the <HELP> key twice and reference H#47…
To get clarification of the reply to my question I follow this – as described above when I press the <HELP> key twice I have to email my query to the Bloomberg Help Desk (and get another reference number).
Posts in praise of the old live help service:
Help! Help Help! and CHEAT: Getting Assistance on Bloomberg (datapoints blog, October 2013)
Bloomberg… real time information on just about everything (Judge Business School Library blog, March 2014)