A Study of Characteristics of Interim Cash Flow Series and Their Proxies and the Ability to Predict Them

Journal of Accounting Advances (J.A.A)
Vol. 5, No. 1, 2013, Ser. 64/3
 
 
Extended Abstract
 
A Study of Characteristics of Interim Cash Flow Series and Their Proxies and the Ability to Predict Them
 
         Dr. M. A. Aghaei              V. Ahmadian             F. Daviran
Tarbiat Modares University
 
Introduction
The predicting of annual and interim cash flows from operations is one of the most important financial information users^, interesting, that is done based on a) reported cash flows from operations or b) the other series as proxies for CFO series. Proxies are usually simplistic algorithms from financial statements’ subcomponents and are used by researchers for the following reasons: 1) these proxies calculate consistently across sample firms (e.g. Bernard & Stober 1989);  2) proxies of algorithms are simple to maintain adequate sample size for empirical testing (e.g. Hopwood & Mckeown 1992; Lorek et al 1993,1996); 3) there are high correlations between reported cash flows from operations and proxies (e.g. Kim & Kross 2005);  and 4) unavailability of  sufficiently long time-series of reported cash flows from operations (e.g. Dechow et al  1998).
 There are numerous financial events (such as acquisition, reclassification, accounting changes, foreign currency translation etc.) that are not included in proxies’ relatively simplistic algorithms. The financial complexities and nuances of such items may cause the time-series properties and predictive ability of reported cash flows from operations to differ from the proxies (Drtina & Largay 1985; Bahnson et al 1996; Heribar & Collins 2002; Mulford & Comiskey 2002; Lorek & Willinger  2008;  Luo 2008 and Cheng  &  Hollie 2008). As a result the external validity of studies using proxies are compromised.
In an effort to mitigate structural change problems, short time-series are used in annual CFO prediction. This action precludes rigorous SAC (sample autocorrelation function (analysis. In fact the purely seasonal characteristics are aggregated and eliminated in annual data masking a potentially important source of autocorrelation. This underscores the need to delineate precisely the time-series properties and predictive ability of reported interim cash flows from operations and proxy series.
 
Research questions and hypothesis
Questions are:
 1) Is there any difference between time-series properties of reported interim cash flows from operations in accordance with Iranian accounting standard no: 2 (after this CFO) and its proxy (after this PCFO)?
2) Is there any significant difference between predictive ability of CFO and PCFO series?
Hypotheses are:
1) There is a difference between time-series properties of CFO and PCFO.
2) The predictive ability of CFO series is significantly higher than PCFO series.
 
 Research method
We have used BOX-Jenkins methodology to considertime-series properties and to assess predictive ability of CFO and PCFO series. At first we computed the firm-specific SAC and SPAC (sample partial autocorrelation function) values of CFO and PCFO data. Then in accordance with the methodology originally popularized by Foster (1977) and also used by Lorek et al (1993, 1996, 2008), firm-specific SAC and SPAC values were summed across sample firms and averaged to obtain values for common structure analysis.
 We used three prediction models and mean absolute percentage error (MAPE) metric to assess predictive ability of CFO and PCFO series. These models are:
 1) Auto regressive (AR) (110)*(000) - CFO_t = CFO_t-1 + Ø ₁ (CFO_t – CFO_t-1) + δ + a_t
2) Random walk (RW), (010)*(000) - CFO_t = δ + CFO_t-1 + a_t
and
3) Interim random walk (IRW), (000)*(010)-CFO_t = δ + CFO_t-2 + a_t
We obtained data from the interim (middle) financial statements of forty five Tehran stock exchange expected firms from the first half of 1379 until the second half of 1387.
Variables are: a) Reported interim cash flows from operations in accordance with Iranian accounting standard no: 2 (CFO) and b) proxy for CFO series derived with below algorithm:
PCFO=operating income before depreciation-the increase in total current assets minus the increase in cash + increase in total current liabilities (used by Hoopwood & Mckeown, 1992).
 
Results
1) Although there are differences among SAC and SPAC values (common structure) of CFO and PCFO series, appropriate models to explain SAC and SPAC behavior of both series are the same. This finding shows that there is no difference between time-series properties of CFO and PCFO series
2) There is no significant difference between predictive ability of CFO and PCFO series.
 
Discussion and conclusion
There are some limitations to predict cash flows from operations, such as unavailability of a sufficiently long time-series of reported CFO. A solution is using proxies for CFO series. Proxies usually are relatively simplistic algorithms from financial statements’ subcomponents and may have different time-series properties and predictive ability, because some complex financial events are eliminated in these algorithms. On the other hand, proxies are usually used in studies, because there is a high correlation between CFO and PCFO data. In this study analysis of SAC and SPAC values (common structure) of CFO and PCFO series show that there are no significant differences between time-series properties and predictive ability of CFO and PCFO series. These findings are consistent with the findings of Dechow et al (1998), Hong Xie (2001) and Kim & Cross (2005) that the proxies could be as reported CFO series.
Our findings contribute to the growing literature on interim cash flow from operations prediction and suggest that while the CFO series are unavailable, the proxy considered in this study could be used.
 
Keywords: Prediction, Interim Operating Cash flows, Time-series, BOX-Jenkins methodology, Operating Cash flows and Proxies for Cash flows from Operation.