The Role of ICT in cash management and forecasting

ICT and Cash Management and
Forecasting
Mark Silins
Moscow
April 2017
Ideally, the IFMIS should be the Primary
Source of data for Forecasting
• Full integration of all possible cashflows into the
IFMIS –
• Daily Bank Reconciliation – increases reliability
regarding the correctness of the cashflow
transactions
• Commitment Control – recording of contracts when
they are legally enforceable – additional useful
information on future cashflows
• Due Date – terms of trade for the government –
provides absolute certainty regarding contractual
payments over a given period
Commitment Controls
• By just capturing commitments for the top 5-10% of payments by value, you
record 90% of the required cashflows future cashflows
• Including indicative dates at the time commitments are recorded would
obviously be a useful addition - including projects with multiple payment points
201X Budget Expenditures by Amount
Grouping by
amount
Below 1,000
Total Valued
%
Cummulat
ive %
Number of
Transactions
Cummulative
%
%
62761970
0.41%
100.00%
293420
63.11%
100.00%
1,000-2,000
56924069
0.37%
99.59%
40213
8.65%
36.89%
2,000-5,000
155344051
1.01%
99.22%
48085
10.34%
28.24%
5,000-10,000
185210557
1.21%
98.20%
26265
5.65%
17.90%
10,000-20,000
285521427
1.86%
97.00%
20243
4.35%
12.25%
20,000-50,000
50,000100,000
100,000250,000
250,0001,000,000
1,000,0005,000,000
Above
5,000,000
549396510
3.59%
95.13%
17323
3.73%
7.89%
549853350
3.59%
91.55%
7893
1.70%
4.17%
885713958
5.78%
87.96%
5729
1.23%
2.47%
2014668642
13.15%
82.18%
4251
0.91%
1.24%
2307871270
15.06%
69.02%
1174
0.25%
0.32%
8266491590
53.96%
53.96%
320
0.07%
0.07%
15319757400
100.00%
464916
100.00%
Total :
Due Date
• Ideally, forecasting should be three months in advance to
allow issuance of securities - but this is a capacity which
can be developed over time – in the shorter term
developing a due date concept can also be useful
• Once goods and services are received you recognize the
accounts payable. The government due date is a set
period after this, typically 30 days.
• You pay on the 30th day, not before and not after
• This ensures a constant stock of 30 days of all upcoming
contractual payments.
• The IFMIS has absolutely perfect information on all
contractual cash outflows for the next 30 days - these are
the cashflows most difficult to forecast
The Use of Models in Liquidity Forecasting
• Annual forecasting for the underlying cash balance and government
liquidity forecasting are different and generally require different
approaches
• Few countries use complex econometrics models for government liquidity
forecasting – simple historical data analysis generally works best,
supplemented by the gathering of regular information from stakeholders
• You can not just build a model where you plug numbers in for government
liquidity forecasting- this Ignores the variability common in short-term
cashflows
• The problem is that the shorter the period the more volatile the flows –
most intricate models fail to provide a useful analytical framework.
Sometimes sophisticated tools hinder rather than help our work
• Thus developing a daily cashflow historical dataset in excel or a database to
help analyze future flows is likely to be the most effectively model
Who has a short term government liquidity model? What is each
country using at present?
Building a Simple Model is Often the
Best Solution
Forecast W1 D1 D2 D3 D4 D5 W1 D6 D7 D8 D9 D10 W2 W3 W4
Actual
D1 D2 D3 D4 D5 W1
Variation
% % % % % %
Forecast W2 D6 D7 D8 D8 D10 W2 D11 D12 D13 D14 D15 W3 W4 W5
Actual
D6 D7 D8 D8 D10 W2
Variation
% % % % % %
6
Linking down-from and up-to
monthly forecasts is also key
Item
Monthly Forecast
Operating Cashflows
Week One
Day 1 For
Day 1 Act
Day 2 For
Day 2 Act
Day 3 For
Day 3 Act
Day 4 For
Day 4 Act
Import Duty
Stamp Duty
Petroleum Taxes
Income Taxes
VAGST
Offshore Registrations
Licenses
Total Tax revenues
Interest Received
Grants from Donors
Special Purpose Accounts
Other non-tax revenue Revenues
0.00
7
Shared Folders for Regular Updating
by Stakeholders is a Useful Strategy
• This is sometimes referred to as a “Community of Practice”
• It allows common files to be shared and updated within a specific
group – for example the Cashflow Operations Committee
• Inputs would be required from key revenue collectors at least
weekly to update specific trends, issues and anomalies. Deadlines
would be set within the committee rules to allow participants
adequate time to review the new datasets and tables prior to the
weekly meeting – excel templates for regular submissions could be
available
• At the weekly meetings the underlying assumptions and proposed
changes would be discussed and consensus reached by the group
• Daily data could also be shared where additional inputs are
required where variations to forecasts exceed certain threshold
percentages
Monitoring the Model is the key
• It is a forecast, and given its short-term nature,
volatility is normal
• So monitoring each day and week is critical. Also
examining whether short-term trends are becoming
long term structural changes is important
• Ensuring forecasting errors are monitored and
investigated is particularly important in any forecast
• This is not rocket science, but it is interesting how
effective regular analysis can be in developing skills to
predict future trends and anomalies