You are on page 1of 50

D E PA R T M E N T O F F I N A N C E

BUREAU OF LOCAL GOVERNMENT FINANCE

Basic Forecasting Tools


www.blgf.gov.ph
Forecasting

• Forecasting is the prediction, projection, or estimate of


some future activity, event, or occurrence based on
information from the past.

• However, LGU revenue forecasting is more than predicting


the future outcome based on past data, it is also providing
logical context based on a mathematical and statistical
model.

BLGF Data Management and Analysis (DAMA) Training


Forecasting: Example
• For example, we see a man standing by a street corner waiting to
cross. Based on this simple observation, we predict that when the
light turns red, he will cross the street.

• We base our forecast on two factors – logic and historical


information.

• This simple act of recognizing patterns and making a logical


assumptions (e.g., the man is rationale) allows us to forecast his
actions.

BLGF Data Management and Analysis (DAMA) Training


Types of Forecasts

• Economic Forecast
• Predict a variety of economic indicators, like money supply,
inflation rates, interest rates, among others.

• Technological Forecasts
• Predict rates of technological progress and innovation.

• Demand Forecast
• Predict the future demand for a company’s products or services

BLGF Data Management and Analysis (DAMA) Training


The Scope of Forecasting Methods
• The range of tools for forecasting is wide. It can be as simple as growth rates
which simply relies on repetition and as complex as regression and other related
mathematical and statistical modelling which tries to identify and quantify the
factors that determine why a variable (e.g., sales) behaves the way it does.

• However, forecasting is a not purely science. Not all factors can be observed
and/or measured. This is when forecasting becomes an art. Other qualitative
methods are then employed – e.g., Delphi method.

• Not all economists and financial analysts are forecasters. But all forecasters
employ statistical methods.

BLGF Data Management and Analysis (DAMA) Training


Why do we forecast?
• We forecast because as local treasurers we need to plan and budget
for both the medium-term (3 to 5 years) and the short-term (1 year).

• Forecasting provides us a plausible scenario on which to build our


revenue and expenditure plans for the medium and short terms.

• Depending on the accuracy and timeliness of our financial


information on hand and the appropriateness of the forecasting
method we employ to generate our forecasts, we can significantly
reduce the uncertainty of having nothing at all to base our plans on
and gives us the confidence we need to make critical policy and
operational decisions.
BLGF Data Management and Analysis (DAMA) Training
Forecasting Methods

Intuitive or ad hoc methods Formal quantitative methods


• Highly judgmental • Also involve extrapolation, but it
• Based on empirical experience is done in a standard way using a
that varies widely. systematic approach.
• Simple and easy to use, but not • Improved forecasting is possible
always as accurate as formal via the application of the right
quantitative methods. method to identify the
relationship between the variable
• Little or no information about to be forecasted and time itself or
the accuracy of the forecast. several other variables.

BLGF Data Management and Analysis (DAMA) Training


Forecasting Methods

UNPREDICTABLE

• Little or no information is available.


• Predicting the discovery of a new, very cheap form of
energy that produces no pollution.
• Predicting the rate of positive behavioral changes in
Philippine politics.

BLGF Data Management and Analysis (DAMA) Training


Forecasting Methods

Forecasting
Methods

Qualitative Quantitative

Time –
Executive Market Sales Force Delphi Explanatory
Series
Opinion Survey Composite Method Forecasting
Models

BLGF Data Management and Analysis (DAMA) Training


Qualitative Forecasting Methods

• Little or no quantitative information is available, but


sufficient qualitative knowledge exists.
• Subjective in nature.
• DO NOT rely heavily on mathematical computations.

Examples:
- Predicting the speed of passage of a set of amendment to the LGC.
- Forecasting the long-term effects of technology on the consumption
of oil.

BLGF Data Management and Analysis (DAMA) Training


Qualitative Forecasting Methods

• A group of managers meet and collectively


Executive Opinion develop a forecast

• Uses interviews and surveys to judge


Market Survey preferences of customer and to assess demand

Sales Force • Each salesperson estimates sales in his or her


Composite region

• Consensus agreement is reached among a group


Delphi Method of experts

BLGF Data Management and Analysis (DAMA) Training


Quantitative Forecasting Methods

• Sufficient quantitative information is available


• Objective in nature.
• RELY heavily on mathematical model.
Example:

- Predicting the continuation of historical patterns such as growth in


GDP or growth in tax revenues.
- Understanding how explanatory variables such as GDP and the
holding of elections affect local revenue sources.

BLGF Data Management and Analysis (DAMA) Training


Conditions for the Application of Quantitative
Forecasting

• Information about the past is available.


• This information can be quantified in the form of numerical
data.
• Assumption of continuity- some aspects of the past pattern
will continue into the future.

BLGF Data Management and Analysis (DAMA) Training


Quantitative Forecasting Techniques

Explanatory
Forecasting

Time Series
Forecasting
BLGF Data Management and Analysis (DAMA) Training
Explanatory Forecasting
(Quantitative Forecasting Techniques)

• Variable to be forecasted exhibits an explanatory


relationship with one or more explanatory variables.
• Purpose of explanatory model is to establish the form of the
relationship and use it to forecast future values of the
forecast variable.
• Any change in inputs will affect the output of the system in a
predictable way assuming the explanatory relationship will
not change.

BLGF Data Management and Analysis (DAMA) Training


Time Series Forecasting
(Quantitative Forecasting Techniques)

• Treats the system as a “black box” and makes no attempt to


discover the factors affecting its behavior.
• Prediction of the future is based on past values of a variable
and/or past errors but not on explanatory variables which
may affect the system.
• Objective is to discover the pattern in the historical data
series and extrapolate that pattern into the future.

BLGF Data Management and Analysis (DAMA) Training


Time Series Analysis

• Time Series: An ordered sequence of values of a variable at


equally spaced time intervals – daily, weekly, monthly,
quarterly, semestral, annual.

• Time series analysis accounts for the fact that data points
taken over time may have an internal structure (such as
autocorrelation, trend or seasonal variation) that should be
accounted for.

BLGF Data Management and Analysis (DAMA) Training


Time Series Models

Naïve • Uses the last period’s actual value as a forecast

Simple Mean (Average) • Uses an average of all past data as a forecast

• Uses an average of a specified number of the most recent observations,


Simple Moving Average with each observation receiving the same emphasis (weight)

• Uses an average of a specified number of the most recent observations,


Weighted Moving Average with each observation receiving a different emphasis (weight)

• A weighted average procedure with weights declining


Exponential Smoothing exponentially as data become older

Trend Projection • Uses the least squares method to fit a straight line to the data

BLGF Data Management and Analysis (DAMA) Training


Naïve Method

• The forecast for next period (period t+1) will be


equal to this period's actual revenues (At).

Example: In this illustration we assume that each year


(beginning with year 2) we made a forecast, then waited to see
the actual revenues every year. We then made a forecast for
the subsequent year, and so on right through to the forecast for
year 6.

BLGF Data Management and Analysis (DAMA) Training


Naïve Method

Actual Revenues Forecast Revenues


Notes
Year (At) (Ft)

There was no prior revenue data on which


1 310 -- to base a forecast for period 1.
From this point forward, these forecasts
2 365 310 were made on a year-by-year basis.
3 395 365
4 415 395 Naïve Method
5 450 415
6 465 450
7 465

BLGF Data Management and Analysis (DAMA) Training


Simple Mean (Average) Method

• The forecast for next period (period t+1) will be


equal to the average of all past historical data.

Example: In the absence of forecast at year 1, it was assumed to


be 300. Simple mean average method will be used for year 2. In
this illustration it is assumed that each year (beginning with year
2) a forecast was made, then waited to see the actual revenue
every year. We then made a forecast for the subsequent year,
and so on right through to the forecast for year 7.

BLGF Data Management and Analysis (DAMA) Training


Simple Mean (Average) Method
Actual Forecast
Revenues Revenues
Year Notes
(At) (Ft)
1 310 300 This forecast was a guess at the beginning.

From this point forward, these forecasts were made on a


2 365 310.000
year-by-year basis using a simple average approach

3 395 337.500 337.5 = (310 + 365)/2


4 415 356.667 356.667 = (310 + 365 + 395)/3
5 450 371.250 371.250 = (310 + 365 + 395 + 415)/4
6 465 387.000 387.000 = (310 + 365 + 395 + 415 + 450)/5
7 400.000 400.000 = (310 + 365 + 395 + 415 + 450 + 465)/6
BLGF Data Management and Analysis (DAMA) Training
Simple Moving Average Method

• The forecast for next period (period t+1) will be equal to


the average of a specified number of the most recent
observations, with each observation receiving the same
emphasis (weight).

Example 1: In this example, a 2-year simple moving average is used.


In the absence of forecast at year 1, it was assumed to be 300. For
year 2, forecast was made using a naïve method (310). Starting year
3, there is already a sufficient data for the 2-year simple moving
average forecast.

BLGF Data Management and Analysis (DAMA) Training


Simple Moving Average Method
Actual Forecast
Revenues Revenues
Year Notes
(At) (Ft)
1 310 300 This forecast was a guess at the beginning.
2 365 310 This forecast was made using a naïve approach.
From this point forward, these forecasts were made on a
year-by-year basis using a 2-yr moving average
3 395 337.500 approach.

337.500 = (310 + 365)/2


4 415 380.000 380.000 = (365 + 395)/2
5 450 405.000 405.000 = (395 + 415)/2
6 465 432.500 432.500 = (415 + 450)/2
7 457.500 457.500 = (450 + 465)/2
BLGF Data Management and Analysis (DAMA) Training
Simple Moving Average Method
Example 2: In this example, a 3-year simple moving average is used. In
the absence of forecast at year 1, it was assumed to be 300. For year 2
and year 3, forecast was made using a naïve method (310 & 365,
respectively). Starting year 4, there is already a sufficient data for the 3-
year simple moving average forecast.

(Solution in the next slide)

BLGF Data Management and Analysis (DAMA) Training


Simple Moving Average Method
Actual Forecast
Year Revenues Revenues Notes
(At) (Ft)
1 310 300 This forecast was a guess at the beginning.
2 365 310 This forecast was made using a naïve approach.

3 395 365 This forecast was made using a naïve approach.


From this point forward, these forecasts were made on
a year-by-year basis using a 3-yr moving average
4 415 356.667 approach.

356.667 = (310 + 365 + 395)/3


5 450 391.667 391.667 = (365 + 395 + 415)/3
6 465 420.000 420.000 = (395 + 415 + 450)/3
7 433.333 433.333 = (415 + 450 + 465)/3
BLGF Data Management and Analysis (DAMA) Training
Simple Moving Average Method

Using more periods in Using fewer periods in


your moving average your moving average
forecasts will result in forecasts will result in
more stability in the more responsiveness in
forecasts. the forecasts

BLGF Data Management and Analysis (DAMA) Training


Weighted Moving Average Method

The forecast for next period (period t+1) will be equal to a


weighted average of a specified number of the most recent
observations.
Example: In this example, a 3-year simple moving average is used. In the
absence of forecast at year 1, it was assumed to be 300. For year 2 and year 3,
forecast was made using a naïve method (310 & 365, respectively). Starting year
4, there is a sufficient data for the 3-year simple moving average forecast.
Beyond that point there is already a sufficient data for the 3-year weighted
moving average forecasts. The weights to be used are as follows: Most recent
year, 0.5; year prior to that, 0.3; year prior to that, 0.2

BLGF Data Management and Analysis (DAMA) Training


Weighted Moving Average Method
Actual Forecast
Year Revenues Revenues Notes
(At) (Ft)
1 310 300 This forecast was a guess at the beginning.
2 365 310 This forecast was made using a naïve approach.
3 395 365 This forecast was made using a naïve approach.
From this point forward, these forecasts were made on
a year-by-year basis using a 3-yr weighted moving
4 415 369.000 average approach.

369.000 = (310*0.2) + (365*0.3) + (395*0.5)


5 450 399.000 399.000 = (365*0.2) + (395*0.3) + (415*0.5)
6 465 428.500 428.500 = (395*0.2) + (415*0.3) + (450*0.5)
7 450.500 450.500 = (415*0.2) + (450*0.3) + (465*0.5)
BLGF Data Management and Analysis (DAMA) Training
Exponential Smoothing Method
The new forecast for next period (period t) will be calculated as follows:

New forecast = Last period’s forecast + α(Last period’s actual data – Last period’s forecast)
(this box contains all you need to know to apply exponential smoothing)
Ft = Ft-1 + (At-1 – Ft-1) (equation 1)
Ft = At-1 + (1-)Ft-1 (alternate equation 1 – a bit more user friendly)
Where  is a smoothing coefficient whose value is between 0 and 1.

Forecast include a portion of every piece of historical data. Furthermore, there will be
different weights placed on these historical values, with older data receiving lower
weights.

BLGF Data Management and Analysis (DAMA) Training


Exponential Smoothing Method

Example: In this example, exponential smoothing method is


used. In the absence of forecast at year 1, it was assumed to
be 300. For each subsequent year, (beginning year 2), forecast
was made using the exponential smoothing model.

BLGF Data Management and Analysis (DAMA) Training


Exponential Smoothing Method
Actual Forecast
Year Revenues Revenues Notes
(At) (Ft)
1 310 300 This was a guess, since there was no prior demand data.
From this point forward, these forecasts were made on a year-by-
2 365 301 year basis using exponential smoothing with =0.1
Ft = At-1 + (1-)Ft-1
3 395 307.4 307.4 = (0.1)(365) + (1-0.1)301

4 415 316.16 316.16 = (0.1)(395) + (1-0.1)307.4

5 450 326.044 326.044 = (0.1)(415) + (1-0.1)316.16

6 465 338.440 338.440 = (0.1)(450) + (1-0.1)326.044

7 351.096 351.096 = (0.1)(465) + (1-0.1)338.440

BLGF Data Management and Analysis (DAMA) Training


Exponential Smoothing Method
Actual Forecast
Year Revenues Revenues
(At) (Ft) Notes
1 310 300 This was a guess, since there was no prior demand data.

From this point forward, these forecasts were made on a year-by-


2 365 302 year basis using exponential smoothing with =0.2
Ft = At-1 + (1-)Ft-1
3 395 314.6 314.6 = (0.2)(365) + (1-0.2)302

4 415 330.68 330.68 = (0.2)(395) + (1-0.2)314.6

5 450 347.544 347.544 = (0.2)(415) + (1-0.2)330.68

6 465 368.035 368.035 = (0.2)(450) + (1-0.2)347.544

7 387.428 387.428 = (0.2)(465) + (1-0.2)368.035

BLGF Data Management and Analysis (DAMA) Training


Trend Projection Method

• This method is a version of the linear regression technique.


Ultimately, the statistical formulas compute a slope for the
trend line (b) and the point where the line crosses the y-axis
(a). This results in the straight line equation: Y = a + bX

where: X represents the values on the horizontal axis (time),


and Y represents the values on the vertical axis (revenue)

BLGF Data Management and Analysis (DAMA) Training


Trend Projection Method
• For demonstration purposes, values of “a” and “b” will be provided as
follows: a = 295; b = 30
• Hence, the equation Y = a + bX will be translated as:
Y = 295 + 30X

This equation can be used to forecast for any year into the future. For example:

Year 7: Forecast = 295 + 30(7) = 505


Year 8: Forecast = 295 + 30(8) = 535
Year 9: Forecast = 295 + 30(9) = 565

BLGF Data Management and Analysis (DAMA) Training


Growth Rates As
Forecasting Tool

www.blgf.gov.ph
What is growth rate?

• Growth Rate is the speed at which a variable (e.g., Business


Tax Collections) is growing.

• Growth Rate is also a type of Percentage, which quantifies


the change of a variable over a period of time.

• Growth Rate then can be understood as “by what part of the


value of a variable in the previous year did that variable
increase to reach its value in the current year”.

BLGF Data Management and Analysis (DAMA) Training


Simple Growth Rate Formula

GR Xt = [(Xt – Xt-1)/Xt-1] x 100

Where:
X = is a variable (e.g., Total Regular Revenues, etc.)
t = is the time or the year (e.g., 2017)
t-1 = is the time of the year less 1 year or 1 year ago
(e.g., 2017 - 1 = 2016)

BLGF Data Management and Analysis (DAMA) Training


Simple Growth Rate: Example
Data:
2015 Annual Regular Income (2015 ARI) = 1,500,000,000
2016 Annual Regular Income (2016 ARI) = 2,000,000,000

Growth Rate for 2016 = [(2016 ARI – 2015 ARI) / 2015 ARI) x 100
= [(2,000,000,000 – 1,500,000,000)/1,500,000,000] x 100
= [(500,000,000)/1,500,000,000.00] x 100
= [0.333] x 100
= 33.3%

BLGF Data Management and Analysis (DAMA) Training


Interpreting Simple Growth Rates

In talking about increasing and decreasing growth rates, one should not be
confused between this and decreases in the values of variables.

Both increasing and decreasing growth rates mean that the value of the
variable in increasing but faster in periods where the growth rate is
increasing and slower in periods where the growth rate is decreasing.

This is different from negative growth where the value is decreasing.

BLGF Data Management and Analysis (DAMA) Training


Average Annual Growth Rates

• Analysts can also take growth rates from a set of years and
get the simple average which is referred to as the Average
Annual Growth Rate (AAGR).

• The AAGR for simple growth rates is simply the sum of all
growth rates in a specific period (e.g., 2014-2016) divided by
the total number of growth rates (e.g., 3).

BLGF Data Management and Analysis (DAMA) Training


Average Annual Growth Rate (AAGR) Formula
AAGR(t..z) = [GRi + GRi+1 + GRi+2 +……GRz] / N
Where:
AAGR(t…n) = Average Annual Growth Rate for a period
t = First year (e.g., 2014)
t +1 = Year 2 (e.g., 2015)
t+2 = Year 3 (e.g., 2016)
z = Final, Last or Terminal Year
GR = Growth Rate
N = Total Number of Growth Rates

BLGF Data Management and Analysis (DAMA) Training


Average Annual Growth Rate – Example
Take the case of the following growth rate information: GR2014 = 24.5%;
GR2015 = 30.5%; and GR2016 = 33.3%. The AAGR is then computed as:

AAGR (2014-2016) = (GR2014 + GR2015 + GR2016) / N


= 24.5% + 30.5% + 33.3% / 3
= 88.3% / 3
= 29.43%

BLGF Data Management and Analysis (DAMA) Training


Interpreting Average Annual Growth Rates

AAGRs allow analysts to summarize growth performance for a


particular time period.

In computing AAGRs, it should be noted that number of


growth rates that can be computed given a particular set of
data will always be 1 less the number of data points or years
with data. This is an important distinction since some analysts
present their AAGR findings as AAGR(2013-2016) which is really a
three-year average since it does not include the growth rate
for 2013 but is a reference to the end points of the data.

BLGF Data Management and Analysis (DAMA) Training


Weighted Average Annual Growth Rates (Weighted AAGR)

• The simple AAGR assumes that the growth rates for each year are of
equal importance and no year’s output or value has substantially
affected the direction of growth. This is why the growth rates are
simply summed up and divided by the total number of growth rates.

• On the other hand, if the analyst’s determines that output in certain


years should have a greater effect on the average growth rate for the
period, they may use the weighted AAGR to address this condition.

BLGF Data Management and Analysis (DAMA) Training


Weighted AAGR Formula
Weighted AAGR(t…t+z)= [(GRt x Xt/∑Xt..t+z) + (GRt+1 x Xt+1/∑Xt..t+z) + (GRt+2 x Xt+2/∑Xt..+t+z)
+ …..+ (GRt+z x Xt+z/∑Xt..t+z)]

Where:
Weighted AAGR(t…t+z)= Average Annual Growth Rate for a period
X = Variable
t = First year (e.g., 2011)
t +1 = Year 2 (e.g., 2012)
t+2 = Year 3 (e.g., 2013)
t+z = Final, Last or Terminal Year
GR = Growth Rate
∑Xt..z = Sum of the values of X or ∑Xt..z = Xt + Xt+1 + Xt+2 + ……+ Xz

BLGF Data Management and Analysis (DAMA) Training


Weighted AAGR: Example
Year Revenues (Php) Growth Rate Weight
2012 923.23 0.17
2013 1149.43 25% 0.21
2014 1,500.00 30% 0.27
2015 2,000.00 33% 0.36
Total 5,572.66 *29% 1.00

Weighted AAGR(t…t+z)=[(GRt x Xt/∑Xt..t+z) + (GRt+1 x Xt+1/∑Xt..t+z) + …..+ (GRt+z x Xt+z/∑Xt..t+z)]


= (25% x 0.21) + (30% x 0.27) + (33% x 0.36) = 25%
Note:
Weight is computed as the share of the revenues per year to the total revenues
*Computed using average annual growth rate.
BLGF Data Management and Analysis (DAMA) Training
Limitations of average annual growth rates
• A key problem of using the average annual growth rate formula and the
weighted average annual growth rate formula is that it does not efficiently
address the volatility in the data or years when the values in the data take a
sudden sharp upward or downward movement.

• A major consequence of not accounting or adjusting for the volatility are


growth rates that, when applied to the first years value, do not result in the
end or terminal year value when the average annual growth rate, simple or
weighted, are applied to it year to year.

BLGF Data Management and Analysis (DAMA) Training


References

• Forecasting and Growth Rates, Mr. Raymund Fabre, EU PFM


2 Technical Assistance Team
• Time Series Analysis: Basic Concepts, Dr. Norman R. Ramos,
EU PFM 2 Technical Assistance Team
• Forecasting Fundamentals

BLGF Data Management and Analysis (DAMA) Training


Thank you!

www.blgf.gov.ph