UNCLASSIFIED<\/p>\n
1<\/p>\n
Paper 1047-2021<\/p>\n
SAS\u00ae Time Series Analysis & Forecasting<\/p>\n
(TSAF) at the Canada Revenue Agency (CRA), with COVID impacts<\/p>\n
Jason A. Oliver, Senior Compliance Analyst, Canada Revenue Agency (CRA)<\/p>\n
ABSTRACT<\/p>\n
It may well be a recurring theme of this year’s SAS Global Forum that we are faced with more pressure to use flexible thinking – not just critical thinking – and when it comes to<\/p>\n
time series analysis and forecasting (TSAF) in SAS, it’s all about “rethinking the curve”.<\/p>\n
At the Canada Revenue Agency (CRA) Compliance Programs Branch (CPB), we have grappled with reliable forecasting for macro-level tax variables on a month-to-month basis, even before the COVID-19 pandemic hit. But now we face a particularly difficult<\/p>\n
challenge. As with many large organizations, it is not easy to foretell what the fallout may be from such a cataclysm.<\/p>\n
In setting up SAS to right the trajectory, we must be extra cautious about some of the fallacies in applying TSAF in this context: the lagged effect for tax revenues realized based on audits of the previous tax year, the need to differentiate average tax recovery<\/p>\n
per case from sum of tax recovery (month-to-month), realizing that industry sectors are not “one size fits all”, and accounting for relatively temporary effects of staffing re-<\/p>\n
orientation in the conversion to a virtual workplace versus the more enduring effects of business disruptions. With SAS Enterprise Miner’s abilities to continuously adjust forecasts, sub-categorize datapoints by tax office or industry sector, and apply lagged<\/p>\n
cross-correlation analysis, we are suitably equipped with the right tools and this can provide abstract learnings for other large organizations.<\/p>\n
INTRODUCTION<\/p>\n
The Canada Revenue Agency (CRA) is Canada\u2019s federal tax administration. As with all tax<\/p>\n
jurisdictions, the CRA has been challenged to keep pace with COVID-19 shocks and<\/p>\n
manifestations, which began in March 2020 (the last month of our fiscal year).<\/p>\n
Fortunately, SAS\u00ae Enterprise Miner\u2122 has been an invaluable aid in gauging these impacts.<\/p>\n
Enterprise Miner\u2122 includes a highly versatile set of functional nodes for configuring and<\/p>\n
processing time series data. It can decompose time series components such as seasonality<\/p>\n
and trend, show trend lines and expected forecast within configurable prediction intervals,<\/p>\n
and demonstrate complex correlation analyses.<\/p>\n
While this has been of great benefit to the CRA in gauging the trajectory of macro-variables<\/p>\n
related to tax revenues and auditor performance, the findings of this research paper could<\/p>\n
UNCLASSIFIED<\/p>\n
2<\/p>\n
conceivably be applied in the abstract to large organizations with process-oriented<\/p>\n
functions, and not just to other foreign tax jurisdictions.<\/p>\n
Let us provide a Glossary of terms to set the stage:<\/p>\n
\uf0b7 TSAF: Time Series Analysis & Forecasting.<\/p>\n
\uf0b7 TEBA: tax earned by audit, which is the amount of tax collectible that is agreed upon in the course of a taxpayer audit. It is in NPV (Net Present Value).<\/p>\n
\uf0b7 TAR: the tax-at-risk, which is the amount that CRA risk assessors arrive at as the precursor to auditing activity.<\/p>\n
\uf0b7 C\/AR ratio: the ratio of [audit] cases completed, to action requests [submitted]<\/p>\n
for assistance. It is a tentative measure of auditor productivity.<\/p>\n
\uf0b7 Integras: the tool used by CRA auditors to process cases.<\/p>\n
TIME SERIES FUNCTIONAL NODES & SETUP<\/p>\n
In SAS\u00ae Enterprise Miner\u2122, you have six TSAF nodes in the \u201cTime Series\u201d ribbon; but we\u2019re<\/p>\n
only going to use four of them. Below is the Time Series ribbon with the functional nodes in<\/p>\n
question:<\/p>\n
Figure 1. Time Series Functional Nodes<\/p>\n
\uf0b7 TS Data Preparation: this node allows you to specify basic time series properties<\/p>\n
including interval, cycle, start\/end time, and accumulation (i.e. by total, min or max,<\/p>\n
mean, etc.)<\/p>\n
o Below, the interval is \u201cautomatic\u201d, so we specify \u201cMonth\u201d as the interval.<\/p>\n
o We can leave the seasonal cycle and start\/end time as \u201cDefault\u201d, as SAS\u00ae<\/p>\n
Enterprise Miner\u2122 will auto-determine these parts from the data.<\/p>\n
o In our case, the data was pre-accumulated in SAS\u00ae Enterprise Guide\u2122 row-<\/p>\n
by-row on a per-month basis, so we can leave Accumulation = \u201cTotal\u201d (else,<\/p>\n
we would have to set it \u201cAverage\u201d).<\/p>\n
Figure 2. TS Data Preparation node \u2013 basic properties<\/p>\n
UNCLASSIFIED<\/p>\n
3<\/p>\n
\uf0b7 TS Decomposition: this node allows you to specify similar basic settings to that of<\/p>\n
the TS Data Prep node, but the Number of Periods can be configured, and moreover,<\/p>\n
you can configure which Export Components you want to display.<\/p>\n
o By default, it will only display \u201cTrend-Cycle\u201d component (=Yes), which is<\/p>\n
generally regarded as the most salient one.<\/p>\n
o However, in our case, we want to view ALL Components, so we would set that<\/p>\n
value to \u201cYes\u201d.<\/p>\n
Figure 3. TS Decomposition node \u2013properties<\/p>\n
TS Correlation: this node allows you to set up your TSA for autocorrelation analysis, or<\/p>\n
alternatively for CCA (Cross-correlation analysis). When you select one of those methods,<\/p>\n
the other one\u2019s properties will be greyed out.<\/p>\n
Figure 4. TS Correlation node \u2013properties<\/p>\n
Both the TS Correlation and TS Decomposition nodes must be preceded by a TS Data<\/p>\n
Preparation node (which occurs right after the source data node).<\/p>\n
UNCLASSIFIED<\/p>\n
4<\/p>\n
TS Exponential Smoothing: this node allows you to conduct forecasting based on your<\/p>\n
known data; as such, you would connect it to a TS Data Preparation node, not directly to<\/p>\n
your source data node.<\/p>\n
\uf0b7 The interval is automatic (which will be month in the case of our pre-accumulated<\/p>\n
data), and the accumulation defaults to \u201cTotal\u201d (which is OK in our case, for the<\/p>\n
same reason).<\/p>\n
\uf0b7 SAS will pick what it deems to be the best forecasting method.<\/p>\n
\uf0b7 The default selection criterion is MSE, or Mean Squared Error.<\/p>\n
\uf0b7 We will see more on the Forecast lead, back, and significance level parameters<\/p>\n
during the forecast demonstration in this paper.<\/p>\n
Figure 5. TS Exponential Smoothing node \u2013properties<\/p>\n
For our initial workspace setup, we can scrutinize on the C\/AR (Case to Action Request)<\/p>\n
ratio, which as per our glossary is a tentative measure of tax auditor performance. The<\/p>\n
initial diagram workspace is called \u201cAggreg_Integras_27mths\u201d, which runs from January<\/p>\n
2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online to March 2020. This is arranged this way for a reason: because it ends on the month<\/p>\n
of the COVID shutdown.<\/p>\n
Our dataset name is \u201cTSA_AGGREG_SINGLE_LINE_27MTHS\u201d.<\/p>\n
So, when I bring this in, I need to set all variables to Role = \u201cRejected\u201d except a) C\/AR ratio<\/p>\n
and b) my MONTH (Time ID) variable.<\/p>\n
Figure 6. Variable Role selection from data source<\/p>\n
UNCLASSIFIED<\/p>\n
5<\/p>\n
You would set your variables once you bring the data source to your diagram (workspace).<\/p>\n
Figure 7. TS Data Source to Diagram flow<\/p>\n
NOTE: I do not cover the mechanics behind bringing in a data source, as the principal focus<\/p>\n
is on conducting TSAF in SAS\u00ae Enterprise Miner\u2122. All we need to be concerned with is that<\/p>\n
as Data Sources become available in the top-left menu, we can drag-and-drop them to our<\/p>\n
diagram workspace (which are also created by right-clicking \u2018Diagrams\u2019 in the left panel).<\/p>\n
In examining the TS Data Preparation node, it is fairly simple: we see the known trajectory of the C\/AR variable, simply by right-clicking the node \uf0e0 Run \uf0e0 Results.<\/p>\n
Figure 8. Time Series Plot, for C\/AR ratio variable<\/p>\n
We can see that the C\/AR ratio has fallen off as of mid-2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online, and continued on a very<\/p>\n
gradual downward path. Which means that case auditors are completing disproportionately<\/p>\n
less cases to the action requests they submit for help, albeit with a seasonal factor and<\/p>\n
some rebounding of the trend-line in March 2020.<\/p>\n
So, we can scrutinize on the more specific components of the time series line by using a TS<\/p>\n
Decomposition node.<\/p>\n
UNCLASSIFIED<\/p>\n
6<\/p>\n
DECOMPOSITION OF TIME SERIES<\/p>\n
In running our TS Decomposition node, and viewing the results, the first one to examine<\/p>\n
is the Seasonal Component Plot. When it comes to the C\/AR ratio, the seasonal index range<\/p>\n
is between a high of about 1.3 down to about 0.75.<\/p>\n
Figure 9. Seasonal Component Plot, for C\/AR ratio variable<\/p>\n
During the months of March and December, we see fairly high seasonality. This is normal<\/p>\n
for the time, since the push to complete cases is higher at the end of the CRA fiscal year<\/p>\n
(March), and ostensibly at the end of the calendar year, also. Auditors are completing<\/p>\n
proportionally more cases vs. the number of action requests they submit to the service<\/p>\n
desk. So it is likely that they are fulfilling cases that do not require as many interventions<\/p>\n
during those months. Even in March 2020, C\/AR still remained high \u2013 it was<\/p>\n
resilient to the initial COVID effects, due to being a ratio variable and not an absolute<\/p>\n
sum variable.<\/p>\n
In the decomposed results, we can also examine combinatory components; for instance, the<\/p>\n
Trend-Cycle Component Plot:<\/p>\n
Figure 10. Trend-Cycle Component Plot, for C\/AR ratio variable<\/p>\n
UNCLASSIFIED<\/p>\n
7<\/p>\n
This tells us what we had surmised from the initial data preparation, that the series has<\/p>\n
been on a steadily downwards trajectory. Now when it comes to tax-related time series<\/p>\n
data, there is no real cycle per se; at best, it is an inherited cycle from world economy<\/p>\n
fluctuations. The proper definition of cycle in a TSA context is not the entity\u2019s operational<\/p>\n
lifecycle; rather, it refers to the boom-and-bust business cycles which are largely<\/p>\n
unpredictable. Ergo, we are mainly concerned about trend here.<\/p>\n
Now, if we substitute the Average TEBA (tax earned by audit) variable for C\/AR [using the<\/p>\n
Data Source node shown in figure 6 earlier], we can see what emerges in our decomposed<\/p>\n
time series results.<\/p>\n
Figure 11. Paneled Component Plots, TS Decomp. for Avg. TEBA<\/p>\n
This time, as per the panel graph at bottom-left, we see that our seasonality index is<\/p>\n
broader than that of C\/AR ratio; it goes from a high of about 1.8 to a low of ~0.7. This is<\/p>\n
largely attributable to the heightened pressures towards fiscal year-end to increase<\/p>\n
realization of TEBA, which we see in Feb.-March. At the opposite end, we see rather low<\/p>\n
seasonality for May, August, and November.<\/p>\n
For the original series plot, bottom-right, the trend continues gradually upwards with<\/p>\n
seasonality readily apparent. In the trend-cycle component plot, at top-left, we see that the<\/p>\n
trend (with cycle, such as it is) is rising steadily upwards but then reaches a virtual plateau.<\/p>\n
The key challenge then, has been to resolve and reconcile the expected forecast as of March<\/p>\n
2020 with the new COVID-19 realities.<\/p>\n
FORECASTING MACRO TAX VARIABLES<\/p>\n
AVERAGE TEBA<\/p>\n
We can proceed to evaluate the expected trajectory of the AVG. TEBA variable, on a<\/p>\n
monthly interval. Recall that this variable is pre-accumulated at data source.<\/p>\n
When we conduct our forecast, we use the TS Exponential Smoothing node.<\/p>\n
UNCLASSIFIED<\/p>\n
8<\/p>\n
Figure 12. TS Exponential Smoothing node in the TSAF diagram<\/p>\n
We let SAS\u00ae pick the best forecasting method, as well as selection criterion (forecast<\/p>\n
measure). In this case, the latter value is the MSE [Mean Squared Error] as you can see at<\/p>\n
the bottom of the properties of the node.<\/p>\n
Figure 13. Properties of the TS Exponential Smoothing node<\/p>\n
For our Significance Level, we set this to 0.5; it governs the blue bracket around the<\/p>\n
forecast line, a.k.a. the prediction interval. So it is a confidence band of sorts. The way this<\/p>\n
figure works is the opposite of what some of us might know from frequentist confidence<\/p>\n
intervals; that is, the lower the \u201calpha\u201d value, the wider the band (prediction interval) so an<\/p>\n
\u201calpha\u201d of 0.01 would produce a very wide band, and an \u201calpha\u201d value = 0.99 would be<\/p>\n
virtually limited to just the forecast line itself. So we aim in the middle (which actually is<\/p>\n
closer to the outline of the trend line, as this figure is more \u201clog-like\u201d in its manifestation).<\/p>\n
Figure 14. TEBA_NPV_Mean: forecast line from trend<\/p>\n
SAS logically expects the trend will continue upwards (while maintaining seasonality, of<\/p>\n
course) due to \u201cseries momentum\u201d. Had we began our time series at, say, January 2016: 2024 – Do my homework – Help write my assignment online<\/p>\n
rather than Jan. 2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online, that momentum might have been more pronounced. The clich\u00e9s of<\/p>\n
UNCLASSIFIED<\/p>\n
9<\/p>\n
\u201cfuture behavior is governed by past behavior\u201d and \u201cyou can\u2019t know where you\u2019re going,<\/p>\n
unless you know where you\u2019ve been\u201d have never been truer. However, enter COVID-19,<\/p>\n
and that is a whole new wrench in the gears of the tax-auditing apparatus.<\/p>\n
As for the selection of \u201cBest\u201d Forecasting Method: you could try to experiment with<\/p>\n
different models \u2013 there are eight in all, as per fundamental TSAF science \u2013 but I can tell<\/p>\n
from the shape of the forecast line that it\u2019s based, appropriately, on the Additive Winters<\/p>\n
method1. I ascertained this by running the node with this method selected, and the<\/p>\n
resulting graph was identical to \u201cbest\u201d method. Unlike the Multiplicative Winters method,<\/p>\n
this forecast line is predicated on fairly consistent seasonal \u201cinverted V\u201d shapes in the curve.<\/p>\n
If those inverted V shapes became noticeable larger (or smaller), then Multiplicative Winters<\/p>\n
would likely be the \u201cbest\u201d method that SAS would auto-select.<\/p>\n
Figure 15. Available Forecasting Methods, properties of TS Exp. Smoothing node<\/p>\n
We see that in the resulting forecast, it predicts ahead exactly 12 months. This is the<\/p>\n
difference between the figures of \u201cForecast Lead\u201d and \u201cForecast Back\u201d in the properties. We<\/p>\n
saw on the previous page that the \u201cForecast Back\u201d = 6; this acts as our validation partition,<\/p>\n
using the last six months of known data (i.e. Oct. 2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers to March 2020). So this gets<\/p>\n
subtracted from the \u201cForecast Back\u201d value of 18 to arrive at 12 periods out. Ideally, you<\/p>\n
want your \u201cback\u201d [validation] period to be between 20-25% of your known data, which it is<\/p>\n
out of 27 months; even when we increase the known months to 30, it will still be 20% of<\/p>\n
this.<\/p>\n
SUM OF TEBA<\/p>\n
When we run a TSAF experiment on the SUM of TEBA \u2013 as opposed to its average \u2013 we<\/p>\n
realize a drastic difference in the scale. Because TEBA is a sum value, not a ratio (i.e.<\/p>\n
C\/AR, or [Average] TEBA\/case), it is simply not as resilient to sudden shocks like COVID-19<\/p>\n
\u2013 as we will later see when adjusting the forecast based on incremental months (April, May,<\/p>\n
June) of known values.<\/p>\n
1 The essence of the Winters method is to combine discernible trend with seasonality.<\/p>\n
UNCLASSIFIED<\/p>\n
10<\/p>\n
Figure 16. TEBA SUM Forecast (post-March 2020)<\/p>\n
Note that the MSE selection criterion (default) graphs a trend line around the known values<\/p>\n
(which are represented by the red dots here). The SUM TEBA for Feb. 2020 is nearly double<\/p>\n
what it was for March 2020, as you can see by the relatively large separation of the red dots<\/p>\n
from the blue dots (on trendline) for those two months. Yet SAS\u00ae \u201cthinks\u201d that the trend<\/p>\n
will continue positively, as it is \u201cCOVID-agnostic\u201d.<\/p>\n
What may also seem shocking to the reader is that the lower limit of the prediction interval<\/p>\n
for April 2020 (at ~$674.5M) actually exceeds the actual value for April 2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers, which was<\/p>\n
slightly below $500 million. It is not until the fall until we see that the midpoint of actual<\/p>\n
2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers data approximates the LCL (lower confidence limit) of the forecasted band for Sept.<\/p>\n
2020. This is ostensibly due to the \u201cpositive momentum\u201d of the time series that I alluded to<\/p>\n
earlier.<\/p>\n
C\/AR RATIO<\/p>\n
Next, we switch out the SUM of TEBA for the C\/AR ratio, once again. In forecasting a<\/p>\n
relatively low continuous ratio variable such as C\/AR, the prediction interval can be less<\/p>\n
reliable. We have to examine the midpoint distribution. While the midpoint post-March<\/p>\n
2020 tends to be at or above the 10.0 line, this is rare for 2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers datapoints.<\/p>\n
Figure 17. C\/AR ratio Forecast<\/p>\n
UNCLASSIFIED<\/p>\n
11<\/p>\n
I used the Mean Relative Abs. Error as the forecast metric (selection criterion), which I<\/p>\n
found to be more appropriate. Regardless, what we see in the actuals for the spring of 2020<\/p>\n
is a very low C\/AR ratio, telling us that case throughput has suffered as a result of the<\/p>\n
pandemic AND that Action Requests for help did not decline proportionally; there was still<\/p>\n
an apparent high need for action requests.<\/p>\n
FORECASTING AVG. HOURS PER CASE<\/p>\n
For forecasting average hours per [audit] case, I determined that the more ideal Selection<\/p>\n
Criterion was \u201cMedian Relative Abs. Error\u201d. No matter what Selection Criterion I used (or<\/p>\n
Significance Level), the prediction interval still dipped into the negative range. Sometimes,<\/p>\n
this is unavoidable. But then the prediction interval becomes spurious; you can\u2019t have<\/p>\n
negative hours. So we tend to just focus on the midpoint values in this situation.<\/p>\n
Figure 18. Average hours per case Forecast<\/p>\n
We can see that the midpoint goes very subtly upwards for the first few forecasted points<\/p>\n
(post-March 2020), then sharply up for summer. As it turns out, this is a fairly good<\/p>\n
approximation of the reality, since the Avg. Hours per case during the middle of 2020 is<\/p>\n
about 1.5-2.0 times that of the previous year. What is especially pronounced is that the<\/p>\n
Average Hours of March 2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers were only 6.25, whereas for March 2020, it was 35.44. This<\/p>\n
was predicated on an Agency policy-induced change; refer to the link and passage below:<\/p>\n
https:\/\/www.mondaq.com\/canada\/audit\/1030308\/cra-moves-forward-with-international-audits- despite-continued-backlog-?email_access=on In March 2020, the CRA announced that it was suspending the vast majority of audit activity for a<\/p>\n
minimum of four weeks, other than audits involving the very largest taxpayers. This suspension meant<\/p>\n
that the CRA ceased requests for information relating to existing audits, finalizing existing audits, and<\/p>\n
issuing reassessments. Further, deadlines for information or document requests were suspended and no<\/p>\n
action was required from taxpayers under audit during this time. This suspension remained in effect until<\/p>\n
June 2020, though audits of small and medium businesses did not resume until late fall.<\/p>\n
This is also arguably responsible for the \u201cpulse\u201d effect we see in actual Avg. TEBA for July<\/p>\n
2020, as per the monthly incremental analysis that comes next.<\/p>\n
https:\/\/www.mondaq.com\/canada\/audit\/1030308\/cra-moves-forward-with-international-audits-despite-continued-backlog-?email_access=on
\nhttps:\/\/www.mondaq.com\/canada\/audit\/1030308\/cra-moves-forward-with-international-audits-despite-continued-backlog-?email_access=on
\nUNCLASSIFIED<\/p>\n
12<\/p>\n
INCREMENTAL ALIGNMENT<\/p>\n
APRIL 2020, KNOWN VALUES<\/p>\n
Now when we add the month of April 2020 to our data (making it 28 mths total), we would<\/p>\n
expect the AVG. TEBA actuals for subsequent months to become closer to \/ within forecast<\/p>\n
range. As an example in the graph cross-section that follows, the forecast for September,<\/p>\n
October, and December 2020 becomes more within range of later-known actuals, once we<\/p>\n
add April 2020 data. However, the July 2020 actual (~$122,000) is still above the forecast<\/p>\n
band for this incremental dataset\u2019s forecast. This was likely due to the resumption of<\/p>\n
standard large business audit as of June 2020 (see previous page article\/passage).<\/p>\n
Figure 19. Revised AVG. TEBA forecast, incremental inclusion of APRIL 2020<\/p>\n
Again, we typically use the measure of MSE [Mean Squared Error] in gauging efficacy or<\/p>\n
proximity of a forecast to actual [values]. See the Appendix tables at the end of this paper<\/p>\n
for a breakdown of this analysis, where I illustrate monthly incremental effect on accuracy<\/p>\n
of the last six months of the calendar year (i.e. from July to Dec. 2020).<\/p>\n
MAY 2020, KNOWN VALUES<\/p>\n
Clearly, the addition of April wasn\u2019t enough to right the trajectory of the expanding \u201cCOVID<\/p>\n
window\u201d. So in continuing our analysis of monthly incremental effect, I added May 2020\u2019s<\/p>\n
known data and I changed the forecast significance level from 0.5 to 0.25. But it makes no<\/p>\n
difference: July actual is still out of forecast range. We must simply accept that July 2020<\/p>\n
Avg. TEBA is an irregular value (~$122K), since July 2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online had Avg. TEBA =~$45K, and July<\/p>\n
2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers\u2019s Avg. TEBA was ~$57K. It is clear that this is a COVID-adjustment spike.<\/p>\n
Figure 20. Revised AVG. TEBA forecast, incremental inclusion of MAY 2020<\/p>\n
UNCLASSIFIED<\/p>\n
13<\/p>\n
We can therefore define July 2020 as a pulse, or a one-time brief event, that caused a<\/p>\n
spike in the accumulated time series value for that month. This emphasis on larger<\/p>\n
business for audit while suspending SMB audits at the time is further substantiated by the<\/p>\n
fact that in July 2020, there was an average of 50.75 hrs per case completed, which is<\/p>\n
extremely high. For April, which had a very high Average TEBA of $185.5K, the figure was<\/p>\n
52.16 average hours per case.<\/p>\n
JUNE 2020, KNOWN VALUES<\/p>\n
Predictably, for the addition of June 2020, it didn\u2019t improve the forecast band to include the<\/p>\n
actual Avg. TEBA for July. So this strengthens the theory that July\u2019s value was a one-time<\/p>\n
event, or pulse, in the time series. It also strengthens the theory that Avg. TEBA was more<\/p>\n
resilient to initial COVID-19 transition measures (being a ratio value, in essence). To wit:<\/p>\n
observe below that the April-May-June line for the original forecast (left) and actual data<\/p>\n
points (right) is just above the $50K line, and follows the same trajectory.<\/p>\n
Figure 21. Comparing Q1 of FY2020-21 forecast vs. actual data points<\/p>\n
In taking MSE and RMSE (R is \u201croot\u201d) measurements for both the as-of-March and as-of-<\/p>\n
June forecasts, we only note a slight improvement (reduction) in that value. Which also<\/p>\n
goes to show the resilience of this variable, and the \u201cpulse\u201d nature of July\u2019s spike.<\/p>\n
MEASURE \/ as of MONTH MARCH 2020 JUNE 2020<\/p>\n
AVG. TEBA (MSE) $ 954,467,257.64 $ 888,454,004.34<\/p>\n
RMSE $ 30,894.45 $ 29,806.95<\/p>\n
Table 1. Point-in-time [R]MSE for AVG. TEBA forecast-to-actual: July to Dec. 2020<\/p>\n
Refer to the Appendix at the end of this paper for a more detailed month-by-month<\/p>\n
breakdown of these calculations.<\/p>\n
FALLACY: COMPARING SUM OF TEBA SHIFT TO AVG. TEBA CHANGES<\/p>\n
TSAF works best when you accumulate data records by average, not by sum total. If we<\/p>\n
tried this exercise using SUM TEBA per month, it would not turn out very well, because sum<\/p>\n
totals are immediately impacted by any severe transition, i.e. auditor work re-arrangements<\/p>\n
and temporary audit case policy due to COVID-19 fallout as of March 2020.<\/p>\n
Evaluating the March 2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers-2020 comparison in the following table, the TEBA_SUM and<\/p>\n
Case Count have dropped significantly in March 2020, yet the C\/AR ratio has augmented.<\/p>\n
UNCLASSIFIED<\/p>\n
14<\/p>\n
Table 2. Year-over-Year March comparison, key macro-variables in TSA<\/p>\n
However, as the staffing situation has attempted to stabilize in the intervening months<\/p>\n
(April to June 2020), the C\/AR ratio has dropped dramatically. (Not shown in above table.)<\/p>\n
The same is true for the TEBA\/AR pattern.<\/p>\n
SUM OF TEBA: DRASTIC CHANGE<\/p>\n
We now compare the SUM TEBA forecast as of March 2020 (left image) and that of June<\/p>\n
2020 known data points (right image).<\/p>\n
Figure 22. Comparison of SUM of TEBA forecast as of March vs. as of June (2020)<\/p>\n
For the first image, none of the actuals of the last six months of 2020 fall in the forecast<\/p>\n
band. Whereas, for the second image, two of the actuals of the last six months (Oct., Nov.)<\/p>\n
fall in the forecast band.<\/p>\n
Also observe how some of the accumulated data points in the forecast are more \u201cdepressed\u201d<\/p>\n
in the latter graph; while there is a discernible peak, it doesn\u2019t quite have the same<\/p>\n
buoyancy or upwards momentum as the former graph. (We must keep in mind, though,<\/p>\n
that this is still using the MSE method, i.e. taking a line of best fit, where the red dots are<\/p>\n
the actual values.)<\/p>\n
So, there is little point in using the MSE to gauge efficacy of the monthly adjustment, simply<\/p>\n
because the values would be so huge (as opposed to those in the Avg. TEBA MSE).<\/p>\n
UNCLASSIFIED<\/p>\n
15<\/p>\n
ADVERSE IMPACTS AND DELAYED EFFECTS<\/p>\n
LATENT EFFECTS OF SHOCKS<\/p>\n
We would also expect that lower Avg. TEBA wouldn\u2019t manifest until much later in the fiscal<\/p>\n
year 2020-21, due to most of 2020 consisting of past year audits. The graph below covers<\/p>\n
known Avg. TEBA trend data points right up to December 2020, the lowest point.<\/p>\n
Figure 23. Calendar-year-end (2020) Avg. TEBA; lowest point<\/p>\n
This extremely low Average TEBA of ~$32,000 per case could be a harbinger of further<\/p>\n
average TEBA decline, but we\u2019d have to observe the last quarter of the fiscal year \u2013 January<\/p>\n
to March 2020, once available \u2013 and validate that theory. (Then we might apply an<\/p>\n
intervention to the time series line.)<\/p>\n
Incidentally, when it comes to SUM of TEBA with actuals up to Dec. 2020, the forecast trend<\/p>\n
line for 2021 is far more credible, showing all datapoints as being well under $1 billion, and<\/p>\n
mostly under $500 million.<\/p>\n
INTERVENTIONS<\/p>\n
As alluded to before, a TSAF exercise may use interventions, if the extreme or irregular<\/p>\n
event is known in advance (or shortly thereafter). This is an adjustment to the \u201cregular\u201d<\/p>\n
time series, using a \u201cdummy\u201d variable for the period of observation. In this case study,<\/p>\n
we\u2019d recommend an intervention for the SUM of TEBA as of March 2020, and possibly for<\/p>\n
AVG TEBA as of Dec. 2020. Plus, we might use a \u201cpulse effect\u201d for July 2020. However,<\/p>\n
programming an intervention requires SAS\u00ae Studio\u2122, which is out of scope for this paper.<\/p>\n
Figure 24. Basic denotation of input variables (interventions) by type<\/p>\n
Lowest actual in 3 years; Dec. 2020 Avg. TEBA of $32,404<\/p>\n
A step would work best as an intervention (for March 2020 and Dec. 2020), since the trend line shift is sudden and sustained; it does not happen gradually then return to baseline.<\/p>\n
UNCLASSIFIED<\/p>\n
16<\/p>\n
TS CORRELATION NODE<\/p>\n
AUTOCORRELATION<\/p>\n
When we deal with a significant seasonal and\/or trend component, we usually find a greater<\/p>\n
degree of autocorrelation factor (abbreviated \u201cACF\u201d). As the name suggests, this is the<\/p>\n
tendency of a variable to self-influence. It could also be regarded as momentum, or<\/p>\n
\u201cmuscle memory\u201d.<\/p>\n
In a similar vein, when frontline auditing teams are performing well, some of that<\/p>\n
momentum carries over from one period to the next, as they build \u201cmuscle memory\u201d and<\/p>\n
are better-equipped to deal with more trying scenarios that have [abstract] aspects in<\/p>\n
common with recent cases worked on. This presents opportunities for \u201cboilerplate\u201d copying<\/p>\n
and pasting of common findings from one case to another, adjusting for specifics, and<\/p>\n
accelerating average time to complete as well as garnering more average TEBA per case.<\/p>\n
Clearly, during the current COVID-19 climate at this writing, and the embargo of SMB case<\/p>\n
audit during the spring 2020 period, we can expect some of that momentum to be adversely<\/p>\n
impacted \u2013 since auditors were working on more complex large business cases overall. But<\/p>\n
first, let us examine a baseline from the years 2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online-2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers, below:<\/p>\n
Figure 25. ACF Plot, three key tax-related macro-variables (2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online-2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers)<\/p>\n
From the three variables plotted above, Est. TAR-AI (tax-at-risk \u2013 audit issue) has low ACF,<\/p>\n
TEBA has moderately high ACF, and Total [Avg. Case] Hours has very high ACF. To wit: at<\/p>\n
lag t=5, TEBA reaches the zero line; but Total Hours is still at ACF=0.45.<\/p>\n
By stark contrast, in 2020 (below), the ACF for both Avg. TEBA and Case Hours is very<\/p>\n
weak overall. In fact, both drop precipitously at the very outset of 2020, just prior to<\/p>\n
COVID-19.<\/p>\n
Figure 26. ACF Plot, same macro-variables, for 2020<\/p>\n
UNCLASSIFIED<\/p>\n
17<\/p>\n
CCA \u2013 CROSS-CORRELATION ANALYSIS<\/p>\n
When we explore lagged effects between risk-related variables \u2013 in this case, TAR (tax-at-<\/p>\n
risk) and TEBA (tax earned by audit) \u2013 we would use a CCA plot. We are also considering<\/p>\n
Total Hours (on audit cases) here. The plots below are at t=3 months and t=12 months<\/p>\n
out, with the influencing variables on the vertical axis, and the influenced variables on the<\/p>\n
X-axis. The color shading is somewhat counterintuitive, whereby red means more positively<\/p>\n
cross-correlated, and blue means less so. Again, we set a baseline of expectations using<\/p>\n
tax data from 2016: 2024 – Do my homework – Help write my assignment online to 2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers (48 months) here.<\/p>\n
Figure 27. CCA Map, at time lags 3 and 12, key macro-variables<\/p>\n
Note the pronounced difference in CCA factor: for time lag 3, the Estimated TAR has<\/p>\n
virtually no effect on TEBA or Total Hours per case (because it\u2019s too close time-wise), but 12<\/p>\n
months out (at right) it has a very pronounced effect on total case hours, and a moderate<\/p>\n
effect on TEBA (~22%). Also, in the first graph for time lag 3, TEBA highly influences Total<\/p>\n
Hours and to a noticeable degree vice-versa too. But when we get to 12 months out, Total<\/p>\n
Hours has virtually no lagged effect on TEBA, and vice-versa.<\/p>\n
If we repeat the experiment from 2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online data up to 2020 (COVID window) data, evaluating<\/p>\n
lagged effects of TAR on TEBA for 2020, we find a very different pattern at t=3 and t=12.<\/p>\n
For time lag=3, the best we get is ~3% influence; for t=12, it\u2019s absolutely nothing.<\/p>\n
Figure 28. CCA Map, at time lags 3 and 12, inclusive of COVID-19 period<\/p>\n
UNCLASSIFIED<\/p>\n
18<\/p>\n
SUBSETTED ANALYSIS<\/p>\n
INDUSTRY PROFILING ANALYSIS<\/p>\n
Using the same data for CCA, we can subdivide our dataset by industry sector, or NAICS<\/p>\n
code. I can set this input to \u201cCross ID\u201d in the data source\u2019s variables list, then re-run the<\/p>\n
flow. From the TS Data Prep node\u2019s Results, right-click in the Time Series Plot and select<\/p>\n
Data Options. We\u2019ll pick a NAICS code at random. And you can see that it fell at the outset<\/p>\n
of COVID, and struggled to regain its footing \u2013 yet exceeding it by calendar year-end.<\/p>\n
Figure 29. Industry Profile (NAICS) subsetting of Avg. TEBA in TS Plot (in 2020)<\/p>\n
Note that when you have over 100 categorical values \u2013 as in the case of NAICS industry<\/p>\n
codes here \u2013 it will only allow you to select from the first 100. In my opinion and<\/p>\n
experience, I prefer SAS VIYA when it comes to subsetting TSA by key categories.<\/p>\n
BY TSO (TAX SERVICES OFFICE)<\/p>\n
So let us examine a subsetting TSA for an under-100 categorical set. I use the TSO, or Tax<\/p>\n
Service Office parameter, so again I set the Case_TSO_ID input to \u201cCross ID\u201d at the data<\/p>\n
source node. Then I re-run the flow and access the Results.<\/p>\n
Figure 30. Tax Services Office (TSO) subsetting of Avg. TEBA in TS Plot (in 2020)<\/p>\n
By default, this will display all TSO IDs in the Input TS Plot; so I have to right-click the plot<\/p>\n
area and select \u201cData Options\u201d to specify filters (WHERE TSO = 5, 18, or 40). Note that<\/p>\n
while all of these TSOs converge at various points, in the month of April we find a very<\/p>\n
strange anomaly: TSO 18 has AVG. TEBA =~ $600K, but the other two TSOs have TEBA<\/p>\n
just under $10,000. Yet all three of them re-converge later in 2020.<\/p>\n
UNCLASSIFIED<\/p>\n
19<\/p>\n
CONCLUSION<\/p>\n
We have seen the power and versatility of SAS\u00ae Enterprise Miner\u2122 for conducting TSAF<\/p>\n
exercises. It is clear that not all macro-variables in the Canada Revenue Agency exhibit the<\/p>\n
same behaviors or resilience at various points in the turbulent COVID-19 period, but a good<\/p>\n
deal of this can be attributed to whether they were pure sum variables, or derived ratio-like<\/p>\n
variables. Some disruptions \u2013 prompting the insertion of intervention effects \u2013 were<\/p>\n
ostensibly due to policies in place to \u201ctake the edge off\u201d more vulnerable business.<\/p>\n
Many of us can also take away abstract learnings from this paper, even if such individuals<\/p>\n
are not employed in the tax sector \u2013 because in the end, it is all about maintaining a certain<\/p>\n
buoyancy of the macro-variables that matter most, to the extent possible \u2013 these are not<\/p>\n
easy times to navigate and we wish those adversely impacted the most clement journey to<\/p>\n
a regained prosperity.<\/p>\n
REFERENCES<\/p>\n
Sarma, Kattamuri S., PhD. Copyright \u00a9 2017. Predictive Modeling with SAS\u00ae Enterprise<\/p>\n
Miner\u2122: Practical Solutions for Business Applications, Third Edition. Cary, NC, USA: SAS<\/p>\n
Institute, Inc.<\/p>\n
ACKNOWLEDGMENTS<\/p>\n
I am grateful to my family for their encouragement on this endeavor. I am also grateful to<\/p>\n
the numerous staff of the CRA who were the audience in my internal presentation of this<\/p>\n
TSAF subject matter. I also acknowledge and admit defeat to the spell checker in insisting<\/p>\n
on the spelling of \u201cendeavor\u201d as it is, not like it ought to be as it is on the space shuttle.<\/p>\n
Which, unlike CRA time series, must be expected to follow a known trajectory.<\/p>\n
RECOMMENDED READING<\/p>\n
\uf0b7 Milh\u00f8j, Anders. Practical Time Series Analysis Using SAS\u00ae. Copyright \u00a9 2013, SAS<\/p>\n
Institute Inc., Cary, NC, USA.<\/p>\n
\uf0b7 Shumway, Robert H. and Stoffer, David S. Time Series Analysis and its Applications. 4th<\/p>\n
ed. \u00a9 Springer International Publishing AG, 2017, Univ. of California at Davis. Davis,<\/p>\n
CA, USA.<\/p>\n
\uf0b7 Brocklebank, John C., Dickey, David A, and Choi, Bong S. SAS\u00ae for Forecasting Time<\/p>\n
Series. 3rd ed. Copyright \u00a9 2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online, SAS Institute Inc., Cary, NC, USA.<\/p>\n
\uf0b7 Svolba, Gerhard. Applying Data Science: Business Case Studies Using SAS\u00ae. Copyright<\/p>\n
\u00a9 2017, SAS Institute Inc., Cary, NC, USA.<\/p>\n
CONTACT INFORMATION<\/p>\n
Your comments and questions are valued and encouraged. Contact the author at:<\/p>\n
Jason A. Oliver, Senior Compliance Analyst & Data Scientist<\/p>\n
Canada Revenue Agency<\/p>\n
Jason.oliver@cra-arc.gc.ca<\/p>\n
mailto:Jason.oliver@cra-arc.gc.ca
\nUNCLASSIFIED<\/p>\n
20<\/p>\n
APPENDIX: TABLES OF ACTUAL-TO-FORECAST ANALYSIS<\/p>\n
This contains detailed breakdowns of the incremental monthly additions of accumulated<\/p>\n
data to the COVID-19 observation window.<\/p>\n
AVERAGE TEBA<\/p>\n
This begins with Average TEBA, being subject to both MSE and RMSE (Mean Squared<\/p>\n
Error, and Root Mean Squared Error).<\/p>\n
At this juncture, between April and May 2020 known data, the MSE \/ RMSE actually<\/p>\n
regresses slightly, telling us that we might as well have gone straight to June 2020\u2019s data.<\/p>\n
In the end, this substantiates our earlier findings, that because Average TEBA is in essence<\/p>\n
a ratio variable and more resilient to initial COVID window \u2013 especially since it is predicated<\/p>\n
UNCLASSIFIED<\/p>\n
21<\/p>\n
on audits of past year\u2019s tax filings \u2013 there was no real near-future benefit to forecast<\/p>\n
alignment based on incremental monthly additions for spring.<\/p>\n
C\/AR RATIO<\/p>\n
This, once again, is the Cases [Completed] to Action Requests [Submitted] ratio. Here I<\/p>\n
break down the monthly forecast measure, using MSE (no RMSE), of the last six months of<\/p>\n
calendar year 2020 and incrementing known months from March up to June. For March to<\/p>\n
May, I include the spring months not yet arrived at in each incremental forecast.<\/p>\n
UNCLASSIFIED<\/p>\n
22<\/p>\n
From adding April known data, the forecast actually worsens; this is arguably due to having<\/p>\n
been accustomed to high C\/AR values for so long. It is not until we add MAY that it becomes<\/p>\n
more realistic.<\/p>\n
Given this extremely low MSE value, brought on by the actual 2.57 C\/AR value of May, we<\/p>\n
have reached the optimum point \u2013 as evidenced by adding June to known values:<\/p>\n
CASE HOURS<\/p>\n
Lastly, in speaking to Hours per [audit] case forecast, I provide a condensed analysis using<\/p>\n
a simplified MAE [Mean Absolute Error] criterion.<\/p>\n
\uf0b7 As of March 2020; forecast of April to Dec. 2020: MAE = 78.52<\/p>\n
\uf0b7 As of April 2020; forecast of May to Dec. 2020: MAE = 95.83<\/p>\n
\uf0b7 As of May 2020; forecast of June to Dec. 2020: MAE = 107.99<\/p>\n
\uf0b7 As of June 2020; forecast of July to Dec. 2020: MAE = 71.51<\/p>\n
So, all in all, this proved a very difficult variable to effectively forecast.<\/p>\n
Applied Sciences
\nArchitecture and Design
\nBiology
\nBusiness & Finance
\nChemistry
\nComputer Science
\nGeography
\nGeology
\nEducation
\nEngineering
\nEnglish
\nEnvironmental science
\nSpanish
\nGovernment
\nHistory
\nHuman Resource Management
\nInformation Systems
\nLaw
\nLiterature
\nMathematics
\nNursing
\nPhysics
\nPolitical Science
\nPsychology
\nReading
\nScience
\nSocial Science
\nHome
\nHomework Answers
\nBlog
\nArchive
\nTags
\nReviews
\nContact
\ntwitterfacebook<\/p>\n
1<\/p>\n
Paper 1047-2021<\/p>\n
SAS\u00ae Time Series Analysis & Forecasting<\/p>\n
(TSAF) at the Canada Revenue Agency (CRA), with COVID impacts<\/p>\n
Jason A. Oliver, Senior Compliance Analyst, Canada Revenue Agency (CRA)<\/p>\n
ABSTRACT<\/p>\n
It may well be a recurring theme of this year’s SAS Global Forum that we are faced with more pressure to use flexible thinking – not just critical thinking – and when it comes to<\/p>\n
time series analysis and forecasting (TSAF) in SAS, it’s all about “rethinking the curve”.<\/p>\n
At the Canada Revenue Agency (CRA) Compliance Programs Branch (CPB), we have grappled with reliable forecasting for macro-level tax variables on a month-to-month basis, even before the COVID-19 pandemic hit. But now we face a particularly difficult<\/p>\n
challenge. As with many large organizations, it is not easy to foretell what the fallout may be from such a cataclysm.<\/p>\n
In setting up SAS to right the trajectory, we must be extra cautious about some of the fallacies in applying TSAF in this context: the lagged effect for tax revenues realized based on audits of the previous tax year, the need to differentiate average tax recovery<\/p>\n
per case from sum of tax recovery (month-to-month), realizing that industry sectors are not “one size fits all”, and accounting for relatively temporary effects of staffing re-<\/p>\n
orientation in the conversion to a virtual workplace versus the more enduring effects of business disruptions. With SAS Enterprise Miner’s abilities to continuously adjust forecasts, sub-categorize datapoints by tax office or industry sector, and apply lagged<\/p>\n
cross-correlation analysis, we are suitably equipped with the right tools and this can provide abstract learnings for other large organizations.<\/p>\n
INTRODUCTION<\/p>\n
The Canada Revenue Agency (CRA) is Canada\u2019s federal tax administration. As with all tax<\/p>\n
jurisdictions, the CRA has been challenged to keep pace with COVID-19 shocks and<\/p>\n
manifestations, which began in March 2020 (the last month of our fiscal year).<\/p>\n
Fortunately, SAS\u00ae Enterprise Miner\u2122 has been an invaluable aid in gauging these impacts.<\/p>\n
Enterprise Miner\u2122 includes a highly versatile set of functional nodes for configuring and<\/p>\n
processing time series data. It can decompose time series components such as seasonality<\/p>\n
and trend, show trend lines and expected forecast within configurable prediction intervals,<\/p>\n
and demonstrate complex correlation analyses.<\/p>\n
While this has been of great benefit to the CRA in gauging the trajectory of macro-variables<\/p>\n
related to tax revenues and auditor performance, the findings of this research paper could<\/p>\n
UNCLASSIFIED<\/p>\n
2<\/p>\n
conceivably be applied in the abstract to large organizations with process-oriented<\/p>\n
functions, and not just to other foreign tax jurisdictions.<\/p>\n
Let us provide a Glossary of terms to set the stage:<\/p>\n
\uf0b7 TSAF: Time Series Analysis & Forecasting.<\/p>\n
\uf0b7 TEBA: tax earned by audit, which is the amount of tax collectible that is agreed upon in the course of a taxpayer audit. It is in NPV (Net Present Value).<\/p>\n
\uf0b7 TAR: the tax-at-risk, which is the amount that CRA risk assessors arrive at as the precursor to auditing activity.<\/p>\n
\uf0b7 C\/AR ratio: the ratio of [audit] cases completed, to action requests [submitted]<\/p>\n
for assistance. It is a tentative measure of auditor productivity.<\/p>\n
\uf0b7 Integras: the tool used by CRA auditors to process cases.<\/p>\n
TIME SERIES FUNCTIONAL NODES & SETUP<\/p>\n
In SAS\u00ae Enterprise Miner\u2122, you have six TSAF nodes in the \u201cTime Series\u201d ribbon; but we\u2019re<\/p>\n
only going to use four of them. Below is the Time Series ribbon with the functional nodes in<\/p>\n
question:<\/p>\n
Figure 1. Time Series Functional Nodes<\/p>\n
\uf0b7 TS Data Preparation: this node allows you to specify basic time series properties<\/p>\n
including interval, cycle, start\/end time, and accumulation (i.e. by total, min or max,<\/p>\n
mean, etc.)<\/p>\n
o Below, the interval is \u201cautomatic\u201d, so we specify \u201cMonth\u201d as the interval.<\/p>\n
o We can leave the seasonal cycle and start\/end time as \u201cDefault\u201d, as SAS\u00ae<\/p>\n
Enterprise Miner\u2122 will auto-determine these parts from the data.<\/p>\n
o In our case, the data was pre-accumulated in SAS\u00ae Enterprise Guide\u2122 row-<\/p>\n
by-row on a per-month basis, so we can leave Accumulation = \u201cTotal\u201d (else,<\/p>\n
we would have to set it \u201cAverage\u201d).<\/p>\n
Figure 2. TS Data Preparation node \u2013 basic properties<\/p>\n
UNCLASSIFIED<\/p>\n
3<\/p>\n
\uf0b7 TS Decomposition: this node allows you to specify similar basic settings to that of<\/p>\n
the TS Data Prep node, but the Number of Periods can be configured, and moreover,<\/p>\n
you can configure which Export Components you want to display.<\/p>\n
o By default, it will only display \u201cTrend-Cycle\u201d component (=Yes), which is<\/p>\n
generally regarded as the most salient one.<\/p>\n
o However, in our case, we want to view ALL Components, so we would set that<\/p>\n
value to \u201cYes\u201d.<\/p>\n
Figure 3. TS Decomposition node \u2013properties<\/p>\n
TS Correlation: this node allows you to set up your TSA for autocorrelation analysis, or<\/p>\n
alternatively for CCA (Cross-correlation analysis). When you select one of those methods,<\/p>\n
the other one\u2019s properties will be greyed out.<\/p>\n
Figure 4. TS Correlation node \u2013properties<\/p>\n
Both the TS Correlation and TS Decomposition nodes must be preceded by a TS Data<\/p>\n
Preparation node (which occurs right after the source data node).<\/p>\n
UNCLASSIFIED<\/p>\n
4<\/p>\n
TS Exponential Smoothing: this node allows you to conduct forecasting based on your<\/p>\n
known data; as such, you would connect it to a TS Data Preparation node, not directly to<\/p>\n
your source data node.<\/p>\n
\uf0b7 The interval is automatic (which will be month in the case of our pre-accumulated<\/p>\n
data), and the accumulation defaults to \u201cTotal\u201d (which is OK in our case, for the<\/p>\n
same reason).<\/p>\n
\uf0b7 SAS will pick what it deems to be the best forecasting method.<\/p>\n
\uf0b7 The default selection criterion is MSE, or Mean Squared Error.<\/p>\n
\uf0b7 We will see more on the Forecast lead, back, and significance level parameters<\/p>\n
during the forecast demonstration in this paper.<\/p>\n
Figure 5. TS Exponential Smoothing node \u2013properties<\/p>\n
For our initial workspace setup, we can scrutinize on the C\/AR (Case to Action Request)<\/p>\n
ratio, which as per our glossary is a tentative measure of tax auditor performance. The<\/p>\n
initial diagram workspace is called \u201cAggreg_Integras_27mths\u201d, which runs from January<\/p>\n
2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online to March 2020. This is arranged this way for a reason: because it ends on the month<\/p>\n
of the COVID shutdown.<\/p>\n
Our dataset name is \u201cTSA_AGGREG_SINGLE_LINE_27MTHS\u201d.<\/p>\n
So, when I bring this in, I need to set all variables to Role = \u201cRejected\u201d except a) C\/AR ratio<\/p>\n
and b) my MONTH (Time ID) variable.<\/p>\n
Figure 6. Variable Role selection from data source<\/p>\n
UNCLASSIFIED<\/p>\n
5<\/p>\n
You would set your variables once you bring the data source to your diagram (workspace).<\/p>\n
Figure 7. TS Data Source to Diagram flow<\/p>\n
NOTE: I do not cover the mechanics behind bringing in a data source, as the principal focus<\/p>\n
is on conducting TSAF in SAS\u00ae Enterprise Miner\u2122. All we need to be concerned with is that<\/p>\n
as Data Sources become available in the top-left menu, we can drag-and-drop them to our<\/p>\n
diagram workspace (which are also created by right-clicking \u2018Diagrams\u2019 in the left panel).<\/p>\n
In examining the TS Data Preparation node, it is fairly simple: we see the known trajectory of the C\/AR variable, simply by right-clicking the node \uf0e0 Run \uf0e0 Results.<\/p>\n
Figure 8. Time Series Plot, for C\/AR ratio variable<\/p>\n
We can see that the C\/AR ratio has fallen off as of mid-2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online, and continued on a very<\/p>\n
gradual downward path. Which means that case auditors are completing disproportionately<\/p>\n
less cases to the action requests they submit for help, albeit with a seasonal factor and<\/p>\n
some rebounding of the trend-line in March 2020.<\/p>\n
So, we can scrutinize on the more specific components of the time series line by using a TS<\/p>\n
Decomposition node.<\/p>\n
UNCLASSIFIED<\/p>\n
6<\/p>\n
DECOMPOSITION OF TIME SERIES<\/p>\n
In running our TS Decomposition node, and viewing the results, the first one to examine<\/p>\n
is the Seasonal Component Plot. When it comes to the C\/AR ratio, the seasonal index range<\/p>\n
is between a high of about 1.3 down to about 0.75.<\/p>\n
Figure 9. Seasonal Component Plot, for C\/AR ratio variable<\/p>\n
During the months of March and December, we see fairly high seasonality. This is normal<\/p>\n
for the time, since the push to complete cases is higher at the end of the CRA fiscal year<\/p>\n
(March), and ostensibly at the end of the calendar year, also. Auditors are completing<\/p>\n
proportionally more cases vs. the number of action requests they submit to the service<\/p>\n
desk. So it is likely that they are fulfilling cases that do not require as many interventions<\/p>\n
during those months. Even in March 2020, C\/AR still remained high \u2013 it was<\/p>\n
resilient to the initial COVID effects, due to being a ratio variable and not an absolute<\/p>\n
sum variable.<\/p>\n
In the decomposed results, we can also examine combinatory components; for instance, the<\/p>\n
Trend-Cycle Component Plot:<\/p>\n
Figure 10. Trend-Cycle Component Plot, for C\/AR ratio variable<\/p>\n
UNCLASSIFIED<\/p>\n
7<\/p>\n
This tells us what we had surmised from the initial data preparation, that the series has<\/p>\n
been on a steadily downwards trajectory. Now when it comes to tax-related time series<\/p>\n
data, there is no real cycle per se; at best, it is an inherited cycle from world economy<\/p>\n
fluctuations. The proper definition of cycle in a TSA context is not the entity\u2019s operational<\/p>\n
lifecycle; rather, it refers to the boom-and-bust business cycles which are largely<\/p>\n
unpredictable. Ergo, we are mainly concerned about trend here.<\/p>\n
Now, if we substitute the Average TEBA (tax earned by audit) variable for C\/AR [using the<\/p>\n
Data Source node shown in figure 6 earlier], we can see what emerges in our decomposed<\/p>\n
time series results.<\/p>\n
Figure 11. Paneled Component Plots, TS Decomp. for Avg. TEBA<\/p>\n
This time, as per the panel graph at bottom-left, we see that our seasonality index is<\/p>\n
broader than that of C\/AR ratio; it goes from a high of about 1.8 to a low of ~0.7. This is<\/p>\n
largely attributable to the heightened pressures towards fiscal year-end to increase<\/p>\n
realization of TEBA, which we see in Feb.-March. At the opposite end, we see rather low<\/p>\n
seasonality for May, August, and November.<\/p>\n
For the original series plot, bottom-right, the trend continues gradually upwards with<\/p>\n
seasonality readily apparent. In the trend-cycle component plot, at top-left, we see that the<\/p>\n
trend (with cycle, such as it is) is rising steadily upwards but then reaches a virtual plateau.<\/p>\n
The key challenge then, has been to resolve and reconcile the expected forecast as of March<\/p>\n
2020 with the new COVID-19 realities.<\/p>\n
FORECASTING MACRO TAX VARIABLES<\/p>\n
AVERAGE TEBA<\/p>\n
We can proceed to evaluate the expected trajectory of the AVG. TEBA variable, on a<\/p>\n
monthly interval. Recall that this variable is pre-accumulated at data source.<\/p>\n
When we conduct our forecast, we use the TS Exponential Smoothing node.<\/p>\n
UNCLASSIFIED<\/p>\n
8<\/p>\n
Figure 12. TS Exponential Smoothing node in the TSAF diagram<\/p>\n
We let SAS\u00ae pick the best forecasting method, as well as selection criterion (forecast<\/p>\n
measure). In this case, the latter value is the MSE [Mean Squared Error] as you can see at<\/p>\n
the bottom of the properties of the node.<\/p>\n
Figure 13. Properties of the TS Exponential Smoothing node<\/p>\n
For our Significance Level, we set this to 0.5; it governs the blue bracket around the<\/p>\n
forecast line, a.k.a. the prediction interval. So it is a confidence band of sorts. The way this<\/p>\n
figure works is the opposite of what some of us might know from frequentist confidence<\/p>\n
intervals; that is, the lower the \u201calpha\u201d value, the wider the band (prediction interval) so an<\/p>\n
\u201calpha\u201d of 0.01 would produce a very wide band, and an \u201calpha\u201d value = 0.99 would be<\/p>\n
virtually limited to just the forecast line itself. So we aim in the middle (which actually is<\/p>\n
closer to the outline of the trend line, as this figure is more \u201clog-like\u201d in its manifestation).<\/p>\n
Figure 14. TEBA_NPV_Mean: forecast line from trend<\/p>\n
SAS logically expects the trend will continue upwards (while maintaining seasonality, of<\/p>\n
course) due to \u201cseries momentum\u201d. Had we began our time series at, say, January 2016: 2024 – Do my homework – Help write my assignment online<\/p>\n
rather than Jan. 2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online, that momentum might have been more pronounced. The clich\u00e9s of<\/p>\n
UNCLASSIFIED<\/p>\n
9<\/p>\n
\u201cfuture behavior is governed by past behavior\u201d and \u201cyou can\u2019t know where you\u2019re going,<\/p>\n
unless you know where you\u2019ve been\u201d have never been truer. However, enter COVID-19,<\/p>\n
and that is a whole new wrench in the gears of the tax-auditing apparatus.<\/p>\n
As for the selection of \u201cBest\u201d Forecasting Method: you could try to experiment with<\/p>\n
different models \u2013 there are eight in all, as per fundamental TSAF science \u2013 but I can tell<\/p>\n
from the shape of the forecast line that it\u2019s based, appropriately, on the Additive Winters<\/p>\n
method1. I ascertained this by running the node with this method selected, and the<\/p>\n
resulting graph was identical to \u201cbest\u201d method. Unlike the Multiplicative Winters method,<\/p>\n
this forecast line is predicated on fairly consistent seasonal \u201cinverted V\u201d shapes in the curve.<\/p>\n
If those inverted V shapes became noticeable larger (or smaller), then Multiplicative Winters<\/p>\n
would likely be the \u201cbest\u201d method that SAS would auto-select.<\/p>\n
Figure 15. Available Forecasting Methods, properties of TS Exp. Smoothing node<\/p>\n
We see that in the resulting forecast, it predicts ahead exactly 12 months. This is the<\/p>\n
difference between the figures of \u201cForecast Lead\u201d and \u201cForecast Back\u201d in the properties. We<\/p>\n
saw on the previous page that the \u201cForecast Back\u201d = 6; this acts as our validation partition,<\/p>\n
using the last six months of known data (i.e. Oct. 2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers to March 2020). So this gets<\/p>\n
subtracted from the \u201cForecast Back\u201d value of 18 to arrive at 12 periods out. Ideally, you<\/p>\n
want your \u201cback\u201d [validation] period to be between 20-25% of your known data, which it is<\/p>\n
out of 27 months; even when we increase the known months to 30, it will still be 20% of<\/p>\n
this.<\/p>\n
SUM OF TEBA<\/p>\n
When we run a TSAF experiment on the SUM of TEBA \u2013 as opposed to its average \u2013 we<\/p>\n
realize a drastic difference in the scale. Because TEBA is a sum value, not a ratio (i.e.<\/p>\n
C\/AR, or [Average] TEBA\/case), it is simply not as resilient to sudden shocks like COVID-19<\/p>\n
\u2013 as we will later see when adjusting the forecast based on incremental months (April, May,<\/p>\n
June) of known values.<\/p>\n
1 The essence of the Winters method is to combine discernible trend with seasonality.<\/p>\n
UNCLASSIFIED<\/p>\n
10<\/p>\n
Figure 16. TEBA SUM Forecast (post-March 2020)<\/p>\n
Note that the MSE selection criterion (default) graphs a trend line around the known values<\/p>\n
(which are represented by the red dots here). The SUM TEBA for Feb. 2020 is nearly double<\/p>\n
what it was for March 2020, as you can see by the relatively large separation of the red dots<\/p>\n
from the blue dots (on trendline) for those two months. Yet SAS\u00ae \u201cthinks\u201d that the trend<\/p>\n
will continue positively, as it is \u201cCOVID-agnostic\u201d.<\/p>\n
What may also seem shocking to the reader is that the lower limit of the prediction interval<\/p>\n
for April 2020 (at ~$674.5M) actually exceeds the actual value for April 2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers, which was<\/p>\n
slightly below $500 million. It is not until the fall until we see that the midpoint of actual<\/p>\n
2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers data approximates the LCL (lower confidence limit) of the forecasted band for Sept.<\/p>\n
2020. This is ostensibly due to the \u201cpositive momentum\u201d of the time series that I alluded to<\/p>\n
earlier.<\/p>\n
C\/AR RATIO<\/p>\n
Next, we switch out the SUM of TEBA for the C\/AR ratio, once again. In forecasting a<\/p>\n
relatively low continuous ratio variable such as C\/AR, the prediction interval can be less<\/p>\n
reliable. We have to examine the midpoint distribution. While the midpoint post-March<\/p>\n
2020 tends to be at or above the 10.0 line, this is rare for 2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers datapoints.<\/p>\n
Figure 17. C\/AR ratio Forecast<\/p>\n
UNCLASSIFIED<\/p>\n
11<\/p>\n
I used the Mean Relative Abs. Error as the forecast metric (selection criterion), which I<\/p>\n
found to be more appropriate. Regardless, what we see in the actuals for the spring of 2020<\/p>\n
is a very low C\/AR ratio, telling us that case throughput has suffered as a result of the<\/p>\n
pandemic AND that Action Requests for help did not decline proportionally; there was still<\/p>\n
an apparent high need for action requests.<\/p>\n
FORECASTING AVG. HOURS PER CASE<\/p>\n
For forecasting average hours per [audit] case, I determined that the more ideal Selection<\/p>\n
Criterion was \u201cMedian Relative Abs. Error\u201d. No matter what Selection Criterion I used (or<\/p>\n
Significance Level), the prediction interval still dipped into the negative range. Sometimes,<\/p>\n
this is unavoidable. But then the prediction interval becomes spurious; you can\u2019t have<\/p>\n
negative hours. So we tend to just focus on the midpoint values in this situation.<\/p>\n
Figure 18. Average hours per case Forecast<\/p>\n
We can see that the midpoint goes very subtly upwards for the first few forecasted points<\/p>\n
(post-March 2020), then sharply up for summer. As it turns out, this is a fairly good<\/p>\n
approximation of the reality, since the Avg. Hours per case during the middle of 2020 is<\/p>\n
about 1.5-2.0 times that of the previous year. What is especially pronounced is that the<\/p>\n
Average Hours of March 2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers were only 6.25, whereas for March 2020, it was 35.44. This<\/p>\n
was predicated on an Agency policy-induced change; refer to the link and passage below:<\/p>\n
https:\/\/www.mondaq.com\/canada\/audit\/1030308\/cra-moves-forward-with-international-audits- despite-continued-backlog-?email_access=on In March 2020, the CRA announced that it was suspending the vast majority of audit activity for a<\/p>\n
minimum of four weeks, other than audits involving the very largest taxpayers. This suspension meant<\/p>\n
that the CRA ceased requests for information relating to existing audits, finalizing existing audits, and<\/p>\n
issuing reassessments. Further, deadlines for information or document requests were suspended and no<\/p>\n
action was required from taxpayers under audit during this time. This suspension remained in effect until<\/p>\n
June 2020, though audits of small and medium businesses did not resume until late fall.<\/p>\n
This is also arguably responsible for the \u201cpulse\u201d effect we see in actual Avg. TEBA for July<\/p>\n
2020, as per the monthly incremental analysis that comes next.<\/p>\n
https:\/\/www.mondaq.com\/canada\/audit\/1030308\/cra-moves-forward-with-international-audits-despite-continued-backlog-?email_access=on
\nhttps:\/\/www.mondaq.com\/canada\/audit\/1030308\/cra-moves-forward-with-international-audits-despite-continued-backlog-?email_access=on
\nUNCLASSIFIED<\/p>\n
12<\/p>\n
INCREMENTAL ALIGNMENT<\/p>\n
APRIL 2020, KNOWN VALUES<\/p>\n
Now when we add the month of April 2020 to our data (making it 28 mths total), we would<\/p>\n
expect the AVG. TEBA actuals for subsequent months to become closer to \/ within forecast<\/p>\n
range. As an example in the graph cross-section that follows, the forecast for September,<\/p>\n
October, and December 2020 becomes more within range of later-known actuals, once we<\/p>\n
add April 2020 data. However, the July 2020 actual (~$122,000) is still above the forecast<\/p>\n
band for this incremental dataset\u2019s forecast. This was likely due to the resumption of<\/p>\n
standard large business audit as of June 2020 (see previous page article\/passage).<\/p>\n
Figure 19. Revised AVG. TEBA forecast, incremental inclusion of APRIL 2020<\/p>\n
Again, we typically use the measure of MSE [Mean Squared Error] in gauging efficacy or<\/p>\n
proximity of a forecast to actual [values]. See the Appendix tables at the end of this paper<\/p>\n
for a breakdown of this analysis, where I illustrate monthly incremental effect on accuracy<\/p>\n
of the last six months of the calendar year (i.e. from July to Dec. 2020).<\/p>\n
MAY 2020, KNOWN VALUES<\/p>\n
Clearly, the addition of April wasn\u2019t enough to right the trajectory of the expanding \u201cCOVID<\/p>\n
window\u201d. So in continuing our analysis of monthly incremental effect, I added May 2020\u2019s<\/p>\n
known data and I changed the forecast significance level from 0.5 to 0.25. But it makes no<\/p>\n
difference: July actual is still out of forecast range. We must simply accept that July 2020<\/p>\n
Avg. TEBA is an irregular value (~$122K), since July 2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online had Avg. TEBA =~$45K, and July<\/p>\n
2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers\u2019s Avg. TEBA was ~$57K. It is clear that this is a COVID-adjustment spike.<\/p>\n
Figure 20. Revised AVG. TEBA forecast, incremental inclusion of MAY 2020<\/p>\n
UNCLASSIFIED<\/p>\n
13<\/p>\n
We can therefore define July 2020 as a pulse, or a one-time brief event, that caused a<\/p>\n
spike in the accumulated time series value for that month. This emphasis on larger<\/p>\n
business for audit while suspending SMB audits at the time is further substantiated by the<\/p>\n
fact that in July 2020, there was an average of 50.75 hrs per case completed, which is<\/p>\n
extremely high. For April, which had a very high Average TEBA of $185.5K, the figure was<\/p>\n
52.16 average hours per case.<\/p>\n
JUNE 2020, KNOWN VALUES<\/p>\n
Predictably, for the addition of June 2020, it didn\u2019t improve the forecast band to include the<\/p>\n
actual Avg. TEBA for July. So this strengthens the theory that July\u2019s value was a one-time<\/p>\n
event, or pulse, in the time series. It also strengthens the theory that Avg. TEBA was more<\/p>\n
resilient to initial COVID-19 transition measures (being a ratio value, in essence). To wit:<\/p>\n
observe below that the April-May-June line for the original forecast (left) and actual data<\/p>\n
points (right) is just above the $50K line, and follows the same trajectory.<\/p>\n
Figure 21. Comparing Q1 of FY2020-21 forecast vs. actual data points<\/p>\n
In taking MSE and RMSE (R is \u201croot\u201d) measurements for both the as-of-March and as-of-<\/p>\n
June forecasts, we only note a slight improvement (reduction) in that value. Which also<\/p>\n
goes to show the resilience of this variable, and the \u201cpulse\u201d nature of July\u2019s spike.<\/p>\n
MEASURE \/ as of MONTH MARCH 2020 JUNE 2020<\/p>\n
AVG. TEBA (MSE) $ 954,467,257.64 $ 888,454,004.34<\/p>\n
RMSE $ 30,894.45 $ 29,806.95<\/p>\n
Table 1. Point-in-time [R]MSE for AVG. TEBA forecast-to-actual: July to Dec. 2020<\/p>\n
Refer to the Appendix at the end of this paper for a more detailed month-by-month<\/p>\n
breakdown of these calculations.<\/p>\n
FALLACY: COMPARING SUM OF TEBA SHIFT TO AVG. TEBA CHANGES<\/p>\n
TSAF works best when you accumulate data records by average, not by sum total. If we<\/p>\n
tried this exercise using SUM TEBA per month, it would not turn out very well, because sum<\/p>\n
totals are immediately impacted by any severe transition, i.e. auditor work re-arrangements<\/p>\n
and temporary audit case policy due to COVID-19 fallout as of March 2020.<\/p>\n
Evaluating the March 2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers-2020 comparison in the following table, the TEBA_SUM and<\/p>\n
Case Count have dropped significantly in March 2020, yet the C\/AR ratio has augmented.<\/p>\n
UNCLASSIFIED<\/p>\n
14<\/p>\n
Table 2. Year-over-Year March comparison, key macro-variables in TSA<\/p>\n
However, as the staffing situation has attempted to stabilize in the intervening months<\/p>\n
(April to June 2020), the C\/AR ratio has dropped dramatically. (Not shown in above table.)<\/p>\n
The same is true for the TEBA\/AR pattern.<\/p>\n
SUM OF TEBA: DRASTIC CHANGE<\/p>\n
We now compare the SUM TEBA forecast as of March 2020 (left image) and that of June<\/p>\n
2020 known data points (right image).<\/p>\n
Figure 22. Comparison of SUM of TEBA forecast as of March vs. as of June (2020)<\/p>\n
For the first image, none of the actuals of the last six months of 2020 fall in the forecast<\/p>\n
band. Whereas, for the second image, two of the actuals of the last six months (Oct., Nov.)<\/p>\n
fall in the forecast band.<\/p>\n
Also observe how some of the accumulated data points in the forecast are more \u201cdepressed\u201d<\/p>\n
in the latter graph; while there is a discernible peak, it doesn\u2019t quite have the same<\/p>\n
buoyancy or upwards momentum as the former graph. (We must keep in mind, though,<\/p>\n
that this is still using the MSE method, i.e. taking a line of best fit, where the red dots are<\/p>\n
the actual values.)<\/p>\n
So, there is little point in using the MSE to gauge efficacy of the monthly adjustment, simply<\/p>\n
because the values would be so huge (as opposed to those in the Avg. TEBA MSE).<\/p>\n
UNCLASSIFIED<\/p>\n
15<\/p>\n
ADVERSE IMPACTS AND DELAYED EFFECTS<\/p>\n
LATENT EFFECTS OF SHOCKS<\/p>\n
We would also expect that lower Avg. TEBA wouldn\u2019t manifest until much later in the fiscal<\/p>\n
year 2020-21, due to most of 2020 consisting of past year audits. The graph below covers<\/p>\n
known Avg. TEBA trend data points right up to December 2020, the lowest point.<\/p>\n
Figure 23. Calendar-year-end (2020) Avg. TEBA; lowest point<\/p>\n
This extremely low Average TEBA of ~$32,000 per case could be a harbinger of further<\/p>\n
average TEBA decline, but we\u2019d have to observe the last quarter of the fiscal year \u2013 January<\/p>\n
to March 2020, once available \u2013 and validate that theory. (Then we might apply an<\/p>\n
intervention to the time series line.)<\/p>\n
Incidentally, when it comes to SUM of TEBA with actuals up to Dec. 2020, the forecast trend<\/p>\n
line for 2021 is far more credible, showing all datapoints as being well under $1 billion, and<\/p>\n
mostly under $500 million.<\/p>\n
INTERVENTIONS<\/p>\n
As alluded to before, a TSAF exercise may use interventions, if the extreme or irregular<\/p>\n
event is known in advance (or shortly thereafter). This is an adjustment to the \u201cregular\u201d<\/p>\n
time series, using a \u201cdummy\u201d variable for the period of observation. In this case study,<\/p>\n
we\u2019d recommend an intervention for the SUM of TEBA as of March 2020, and possibly for<\/p>\n
AVG TEBA as of Dec. 2020. Plus, we might use a \u201cpulse effect\u201d for July 2020. However,<\/p>\n
programming an intervention requires SAS\u00ae Studio\u2122, which is out of scope for this paper.<\/p>\n
Figure 24. Basic denotation of input variables (interventions) by type<\/p>\n
Lowest actual in 3 years; Dec. 2020 Avg. TEBA of $32,404<\/p>\n
A step would work best as an intervention (for March 2020 and Dec. 2020), since the trend line shift is sudden and sustained; it does not happen gradually then return to baseline.<\/p>\n
UNCLASSIFIED<\/p>\n
16<\/p>\n
TS CORRELATION NODE<\/p>\n
AUTOCORRELATION<\/p>\n
When we deal with a significant seasonal and\/or trend component, we usually find a greater<\/p>\n
degree of autocorrelation factor (abbreviated \u201cACF\u201d). As the name suggests, this is the<\/p>\n
tendency of a variable to self-influence. It could also be regarded as momentum, or<\/p>\n
\u201cmuscle memory\u201d.<\/p>\n
In a similar vein, when frontline auditing teams are performing well, some of that<\/p>\n
momentum carries over from one period to the next, as they build \u201cmuscle memory\u201d and<\/p>\n
are better-equipped to deal with more trying scenarios that have [abstract] aspects in<\/p>\n
common with recent cases worked on. This presents opportunities for \u201cboilerplate\u201d copying<\/p>\n
and pasting of common findings from one case to another, adjusting for specifics, and<\/p>\n
accelerating average time to complete as well as garnering more average TEBA per case.<\/p>\n
Clearly, during the current COVID-19 climate at this writing, and the embargo of SMB case<\/p>\n
audit during the spring 2020 period, we can expect some of that momentum to be adversely<\/p>\n
impacted \u2013 since auditors were working on more complex large business cases overall. But<\/p>\n
first, let us examine a baseline from the years 2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online-2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers, below:<\/p>\n
Figure 25. ACF Plot, three key tax-related macro-variables (2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online-2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers)<\/p>\n
From the three variables plotted above, Est. TAR-AI (tax-at-risk \u2013 audit issue) has low ACF,<\/p>\n
TEBA has moderately high ACF, and Total [Avg. Case] Hours has very high ACF. To wit: at<\/p>\n
lag t=5, TEBA reaches the zero line; but Total Hours is still at ACF=0.45.<\/p>\n
By stark contrast, in 2020 (below), the ACF for both Avg. TEBA and Case Hours is very<\/p>\n
weak overall. In fact, both drop precipitously at the very outset of 2020, just prior to<\/p>\n
COVID-19.<\/p>\n
Figure 26. ACF Plot, same macro-variables, for 2020<\/p>\n
UNCLASSIFIED<\/p>\n
17<\/p>\n
CCA \u2013 CROSS-CORRELATION ANALYSIS<\/p>\n
When we explore lagged effects between risk-related variables \u2013 in this case, TAR (tax-at-<\/p>\n
risk) and TEBA (tax earned by audit) \u2013 we would use a CCA plot. We are also considering<\/p>\n
Total Hours (on audit cases) here. The plots below are at t=3 months and t=12 months<\/p>\n
out, with the influencing variables on the vertical axis, and the influenced variables on the<\/p>\n
X-axis. The color shading is somewhat counterintuitive, whereby red means more positively<\/p>\n
cross-correlated, and blue means less so. Again, we set a baseline of expectations using<\/p>\n
tax data from 2016: 2024 – Do my homework – Help write my assignment online to 2019: 2024 – Online Assignment Homework Writing Help Service By Expert Research Writers (48 months) here.<\/p>\n
Figure 27. CCA Map, at time lags 3 and 12, key macro-variables<\/p>\n
Note the pronounced difference in CCA factor: for time lag 3, the Estimated TAR has<\/p>\n
virtually no effect on TEBA or Total Hours per case (because it\u2019s too close time-wise), but 12<\/p>\n
months out (at right) it has a very pronounced effect on total case hours, and a moderate<\/p>\n
effect on TEBA (~22%). Also, in the first graph for time lag 3, TEBA highly influences Total<\/p>\n
Hours and to a noticeable degree vice-versa too. But when we get to 12 months out, Total<\/p>\n
Hours has virtually no lagged effect on TEBA, and vice-versa.<\/p>\n
If we repeat the experiment from 2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online data up to 2020 (COVID window) data, evaluating<\/p>\n
lagged effects of TAR on TEBA for 2020, we find a very different pattern at t=3 and t=12.<\/p>\n
For time lag=3, the best we get is ~3% influence; for t=12, it\u2019s absolutely nothing.<\/p>\n
Figure 28. CCA Map, at time lags 3 and 12, inclusive of COVID-19 period<\/p>\n
UNCLASSIFIED<\/p>\n
18<\/p>\n
SUBSETTED ANALYSIS<\/p>\n
INDUSTRY PROFILING ANALYSIS<\/p>\n
Using the same data for CCA, we can subdivide our dataset by industry sector, or NAICS<\/p>\n
code. I can set this input to \u201cCross ID\u201d in the data source\u2019s variables list, then re-run the<\/p>\n
flow. From the TS Data Prep node\u2019s Results, right-click in the Time Series Plot and select<\/p>\n
Data Options. We\u2019ll pick a NAICS code at random. And you can see that it fell at the outset<\/p>\n
of COVID, and struggled to regain its footing \u2013 yet exceeding it by calendar year-end.<\/p>\n
Figure 29. Industry Profile (NAICS) subsetting of Avg. TEBA in TS Plot (in 2020)<\/p>\n
Note that when you have over 100 categorical values \u2013 as in the case of NAICS industry<\/p>\n
codes here \u2013 it will only allow you to select from the first 100. In my opinion and<\/p>\n
experience, I prefer SAS VIYA when it comes to subsetting TSA by key categories.<\/p>\n
BY TSO (TAX SERVICES OFFICE)<\/p>\n
So let us examine a subsetting TSA for an under-100 categorical set. I use the TSO, or Tax<\/p>\n
Service Office parameter, so again I set the Case_TSO_ID input to \u201cCross ID\u201d at the data<\/p>\n
source node. Then I re-run the flow and access the Results.<\/p>\n
Figure 30. Tax Services Office (TSO) subsetting of Avg. TEBA in TS Plot (in 2020)<\/p>\n
By default, this will display all TSO IDs in the Input TS Plot; so I have to right-click the plot<\/p>\n
area and select \u201cData Options\u201d to specify filters (WHERE TSO = 5, 18, or 40). Note that<\/p>\n
while all of these TSOs converge at various points, in the month of April we find a very<\/p>\n
strange anomaly: TSO 18 has AVG. TEBA =~ $600K, but the other two TSOs have TEBA<\/p>\n
just under $10,000. Yet all three of them re-converge later in 2020.<\/p>\n
UNCLASSIFIED<\/p>\n
19<\/p>\n
CONCLUSION<\/p>\n
We have seen the power and versatility of SAS\u00ae Enterprise Miner\u2122 for conducting TSAF<\/p>\n
exercises. It is clear that not all macro-variables in the Canada Revenue Agency exhibit the<\/p>\n
same behaviors or resilience at various points in the turbulent COVID-19 period, but a good<\/p>\n
deal of this can be attributed to whether they were pure sum variables, or derived ratio-like<\/p>\n
variables. Some disruptions \u2013 prompting the insertion of intervention effects \u2013 were<\/p>\n
ostensibly due to policies in place to \u201ctake the edge off\u201d more vulnerable business.<\/p>\n
Many of us can also take away abstract learnings from this paper, even if such individuals<\/p>\n
are not employed in the tax sector \u2013 because in the end, it is all about maintaining a certain<\/p>\n
buoyancy of the macro-variables that matter most, to the extent possible \u2013 these are not<\/p>\n
easy times to navigate and we wish those adversely impacted the most clement journey to<\/p>\n
a regained prosperity.<\/p>\n
REFERENCES<\/p>\n
Sarma, Kattamuri S., PhD. Copyright \u00a9 2017. Predictive Modeling with SAS\u00ae Enterprise<\/p>\n
Miner\u2122: Practical Solutions for Business Applications, Third Edition. Cary, NC, USA: SAS<\/p>\n
Institute, Inc.<\/p>\n
ACKNOWLEDGMENTS<\/p>\n
I am grateful to my family for their encouragement on this endeavor. I am also grateful to<\/p>\n
the numerous staff of the CRA who were the audience in my internal presentation of this<\/p>\n
TSAF subject matter. I also acknowledge and admit defeat to the spell checker in insisting<\/p>\n
on the spelling of \u201cendeavor\u201d as it is, not like it ought to be as it is on the space shuttle.<\/p>\n
Which, unlike CRA time series, must be expected to follow a known trajectory.<\/p>\n
RECOMMENDED READING<\/p>\n
\uf0b7 Milh\u00f8j, Anders. Practical Time Series Analysis Using SAS\u00ae. Copyright \u00a9 2013, SAS<\/p>\n
Institute Inc., Cary, NC, USA.<\/p>\n
\uf0b7 Shumway, Robert H. and Stoffer, David S. Time Series Analysis and its Applications. 4th<\/p>\n
ed. \u00a9 Springer International Publishing AG, 2017, Univ. of California at Davis. Davis,<\/p>\n
CA, USA.<\/p>\n
\uf0b7 Brocklebank, John C., Dickey, David A, and Choi, Bong S. SAS\u00ae for Forecasting Time<\/p>\n
Series. 3rd ed. Copyright \u00a9 2018: 2024 – Write My Essay For Me | Essay Writing Service For Your Papers Online, SAS Institute Inc., Cary, NC, USA.<\/p>\n
\uf0b7 Svolba, Gerhard. Applying Data Science: Business Case Studies Using SAS\u00ae. Copyright<\/p>\n
\u00a9 2017, SAS Institute Inc., Cary, NC, USA.<\/p>\n
CONTACT INFORMATION<\/p>\n
Your comments and questions are valued and encouraged. Contact the author at:<\/p>\n
Jason A. Oliver, Senior Compliance Analyst & Data Scientist<\/p>\n
Canada Revenue Agency<\/p>\n
Jason.oliver@cra-arc.gc.ca<\/p>\n
mailto:Jason.oliver@cra-arc.gc.ca
\nUNCLASSIFIED<\/p>\n
20<\/p>\n
APPENDIX: TABLES OF ACTUAL-TO-FORECAST ANALYSIS<\/p>\n
This contains detailed breakdowns of the incremental monthly additions of accumulated<\/p>\n
data to the COVID-19 observation window.<\/p>\n
AVERAGE TEBA<\/p>\n
This begins with Average TEBA, being subject to both MSE and RMSE (Mean Squared<\/p>\n
Error, and Root Mean Squared Error).<\/p>\n
At this juncture, between April and May 2020 known data, the MSE \/ RMSE actually<\/p>\n
regresses slightly, telling us that we might as well have gone straight to June 2020\u2019s data.<\/p>\n
In the end, this substantiates our earlier findings, that because Average TEBA is in essence<\/p>\n
a ratio variable and more resilient to initial COVID window \u2013 especially since it is predicated<\/p>\n
UNCLASSIFIED<\/p>\n
21<\/p>\n
on audits of past year\u2019s tax filings \u2013 there was no real near-future benefit to forecast<\/p>\n
alignment based on incremental monthly additions for spring.<\/p>\n
C\/AR RATIO<\/p>\n
This, once again, is the Cases [Completed] to Action Requests [Submitted] ratio. Here I<\/p>\n
break down the monthly forecast measure, using MSE (no RMSE), of the last six months of<\/p>\n
calendar year 2020 and incrementing known months from March up to June. For March to<\/p>\n
May, I include the spring months not yet arrived at in each incremental forecast.<\/p>\n
UNCLASSIFIED<\/p>\n
22<\/p>\n
From adding April known data, the forecast actually worsens; this is arguably due to having<\/p>\n
been accustomed to high C\/AR values for so long. It is not until we add MAY that it becomes<\/p>\n
more realistic.<\/p>\n
Given this extremely low MSE value, brought on by the actual 2.57 C\/AR value of May, we<\/p>\n
have reached the optimum point \u2013 as evidenced by adding June to known values:<\/p>\n
CASE HOURS<\/p>\n
Lastly, in speaking to Hours per [audit] case forecast, I provide a condensed analysis using<\/p>\n
a simplified MAE [Mean Absolute Error] criterion.<\/p>\n
\uf0b7 As of March 2020; forecast of April to Dec. 2020: MAE = 78.52<\/p>\n
\uf0b7 As of April 2020; forecast of May to Dec. 2020: MAE = 95.83<\/p>\n
\uf0b7 As of May 2020; forecast of June to Dec. 2020: MAE = 107.99<\/p>\n
\uf0b7 As of June 2020; forecast of July to Dec. 2020: MAE = 71.51<\/p>\n
So, all in all, this proved a very difficult variable to effectively forecast.<\/p>\n
Applied Sciences
\nArchitecture and Design
\nBiology
\nBusiness & Finance
\nChemistry
\nComputer Science
\nGeography
\nGeology
\nEducation
\nEngineering
\nEnglish
\nEnvironmental science
\nSpanish
\nGovernment
\nHistory
\nHuman Resource Management
\nInformation Systems
\nLaw
\nLiterature
\nMathematics
\nNursing
\nPhysics
\nPolitical Science
\nPsychology
\nReading
\nScience
\nSocial Science
\nHome
\nHomework Answers
\nBlog
\nArchive
\nTags
\nReviews
\nContact
\ntwitterfacebook<\/p>\n","protected":false},"excerpt":{"rendered":"
UNCLASSIFIED 1 Paper 1047-2021 SAS\u00ae Time Series Analysis & Forecasting (TSAF) at the Canada Revenue Agency (CRA), with COVID impacts Jason A. Oliver, Senior Compliance Analyst, Canada Revenue…<\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1348,1345,92,2503,2494,1347],"tags":[2453,2489,2487,2368,2507,2370,1462,2488,2506,2490,2504,2505],"class_list":["post-8924","post","type-post","status-publish","format-standard","hentry","category-best-nursing-assignment-writing-service","category-cheap-online-nursing-essays-writing","category-nursing","category-nursing-essay-writers","category-online-nursing-essay-writing-service","category-write-my-nursing-paper-online","tag-bsn-papers","tag-dnp-assignment","tag-health-care-essays","tag-masters-essays","tag-nurs-essays","tag-nursing-assessment","tag-nursing-assignment","tag-nursing-homework","tag-online-nursing-essays","tag-online-nursing-papers","tag-professional-nursing-essay-writing","tag-top-nursing-papers"],"_links":{"self":[{"href":"https:\/\/www.homeworkacetutors.com\/nursing\/wp-json\/wp\/v2\/posts\/8924","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.homeworkacetutors.com\/nursing\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.homeworkacetutors.com\/nursing\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.homeworkacetutors.com\/nursing\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/www.homeworkacetutors.com\/nursing\/wp-json\/wp\/v2\/comments?post=8924"}],"version-history":[{"count":0,"href":"https:\/\/www.homeworkacetutors.com\/nursing\/wp-json\/wp\/v2\/posts\/8924\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.homeworkacetutors.com\/nursing\/wp-json\/wp\/v2\/media?parent=8924"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.homeworkacetutors.com\/nursing\/wp-json\/wp\/v2\/categories?post=8924"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.homeworkacetutors.com\/nursing\/wp-json\/wp\/v2\/tags?post=8924"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}