Accrual Ratios are an important index for assessing the performance and continued viability of every publicly traded company. Using Wakari, TR CONNECT, and Thomson Reuters Financial Data I calculated accrual ratios for a variety of companies and explored news events, which may correlate with strong signals in the data. You can easily download and edit the notebook used in this post into your Wakari account.

A standard way to calculate earnings management and quality is to look at accrual ratios. This is essentially a measure of earnings quality and can often be a good indicator for investment and, possibly, fraudulent manipulation. A company that posts high accrual ratios for an extended periods of time warrant further investigation. High ratios can imply that a company is relying more on unrealized future transactions than collected earnings. Because companies are encouraged to show growth each quarter, management may move more finances into Total Assets, which drives the accrual ratio up. Because the economy is constantly in flux, the accrual ratio should similarly have some fluctuations. If a company is reporting similar accrual ratios quarter after quarter, this may indicate a problem with financial management and one should further investigate before investing/divesting.

Computing the accrual ratio is quite simple:




Where NOA (Net Operating Assets) is Total Assets – Cash – (Total Liabilities – Total Debt). These metrics are reported according to standard GAAP practices in the balance sheet of the company each quarter. Thomson Reuters records and makes available all these and many more metrics in their Worldscope Fundamentals database.

Getting the Data and Calculating

As I previously posted, Continuum and Thomson Reuters have partnered to bring QA Direct (of which Worldscope is a part) to Wakari customers. I used Continuum’s TR_CONNECT (a Python interface to QA Direct) to easily extract the necessary data out of Thomson Reuters’s database. I need the following:

  • 2001: Cash
  • 3351: Total Liabilities
  • 3255: Total Debt
  • 2999: Total Assets

where the numerical value is the call signature in the Worldscope Database. For example, to request IBM’s Total Debt, we could write a query like this:

total_debt = tr.query('IBM','ws.3255','Q')



Like all data, financial data is messy. For instance, Apple reports their Total Debt* on the 25th of December and Total Liabilities** on the 26th. This data-alignment problem is fairly common in financial timeseries data. Since TR Connect returns Pandas objects, we can use the resample and set_index function to ensure the calculations are properly date-aligned.

cash =  tr.query('IBM','ws.2001','Q')
cash.df = cash.df.resample('Q',fill_method='ffill')




Let’s look at IBM’s accrual ratio:

There’s a large valley between 2008 and 2010. Let’s grab all of our metrics in one DataFrame and print out the values.

accruals = tr.query(sec,['ws.3255','ws.3351','ws.2001','ws.2999'],freq)
accruals.df.columns = ['total_debt','total_liabilities','Cash','total_assets']
accurals.df = accurals.df.resample('Q',fill_method='ffill')
print accruals.df.ix['2008':'2010']




  total_debt total_liabilities Cash total_assets
2008-03-31 35.2 93.1 12.0 121.8
2008-06-30 34.2 92.7 9.8 120.9
2008-09-30 34.4 88.4 9.7 115.9
2008-12-31 33.9 88.8 13.7 102.3
2009-03-31 27.4 81.7 12.9 954.2
2009-06-30 25.4 81.4 12.7 96.9
2009-09-30 21.3 78.8 11.5 97.3
2009-12-31 26.1 82.1 14.2 104.8
2010-03-31 23.4 79.5 14.3 101.7
2010-06-30 24.3 79.1 13.1 100.3
2010-09-30 24.9 81.9 11.6 104.2
2010-12-31 28.6 87.1 12.1 110.2

values listed in billions


The Total Assets began to creep down starting in the first quarter of 2009 and did not return until the 4th quarter. In July of 2008 IBM sold off much of their shares in Lenovo. This could account for Total Assets moving down and Cash increasing.

Apple (AAP) has posted record profits the past couple of years. What does their accrual ratio over time look like?

Apple’s accrual ratio has a distinctive pulse between 2006 and 2007. Again, let’s inspect the metrics individually.


  total_assets cash total_liabilities total_debt
2005-03-26 10023 7057 3637 0
2005-06-25 10400 7526 3579 0
2005-09-24 11368 8261 3902 0
2005-12-31 14059 8707 5679 0
2006-04-01 13777 8226 5095 0
2006-07-01 15114 9176 5784 0
2006-09-30 17205 10110 7221 0
2006-12-30 19413 11869 8185 0
2007-03-31 18684 12577 6423 0
2007-06-30 21618 13767 8214 0
2007-09-29 25259 15386 10727 0
2007-12-29 29849 18448 13045 0

values listed in millions


It’s a little hard to discern what’s happening here. Potting all the values can shed some light on what happened:

accruals = tr.query('AAPL',['ws.3255','ws.3351','ws.2001','ws.2999'],'Q')



In 2005, all but one metric began to increase. Surprisingly, Total Debt remained 0 throughout the pulse and remains so today! It is still quite difficult to fully understand what is causing the puele. Fortunately, Apple (and many companies) maintain web accessible archives of their quarterly reports. In April of 2005, Apple announced its revenue increased 70% and in June announced they would begin the transition from PowerPC to Intel. It would make sense then that consumers would slowly stop purchasing the PowerPC machines and request orders for the new Intel machines. Apple was relying more on the accrued and future sales than the sales which had already taken place.

In its first quarterly report of 2006 Apple announced the highest revenue and earnings in the Company’s history. During this time, Apple also highlighted new partnerships and additions to the iTunes music catalogue in their monthly statements. Throughout 2006, Apple posted record profits and Steve Jobs noted, “Our transition to Intel processors is going very well, and our music business just experienced another quarter of outstanding growth.” The valley may be a result from a windfall of cash (over $10 Billion at that time) from increased sales of their new Intel machines and the success of iTunes.

Thomson Reuters has generously provided a complete teaser data set for 34 securities exclusively on Wakari and free for our users. Using TR CONNECT you can easily start pulling data on a variety of metrics. If you have any insight into what’s happening I encourage you to clone my notebook and start iterating. If you don’t already have Thomson Reuters credentials and are interested in exploring the data please sign up for Wakari and contact Continuum at [email protected] for access to the teaser data set.


Edit Notebook in Wakari


To running this notebook please contact [email protected]