Iconometric modelling, forecasts and strategies

Iconometric modelling, forecasts and strategies: How does the measurable outcome change, what does it depend on, and what can be expected?

The future is not an observed fact. It cannot be assessed with the same certainty with which we measure actual sales, expenditure, employment or budget revenue. But this does not mean that every decision about the future must be based on intuition. When sufficient data is available, relationships can be modelled, probable developments assessed, and strategic alternatives compared under common and verifiable assumptions.

SOLISFERADOME ANALYTICS develops econometric models, economic forecasts, scenarios and strategies for companies, public institutions, institutional and professional investors, and organisations. The service supports decision-making regarding demand, sales, production, capacity, prices, costs, budgets, revenue, employment, investment, public policy and risk management.

The aim is not to present a single figure as a certainty for the future. The aim is to determine what is likely to happen under certain conditions, which factors drive the outcome, how great the uncertainty is, under what deviations the conclusion ceases to be valid, and what strategy is justified given the client’s objectives and constraints.

What is econometric modelling?

Econometric modelling is the formalised study of economic relationships using economic theory, statistical data and mathematical methods. The model links an outcome that we wish to explain or forecast to factors that we have reason to expect are related to it.

For example, the demand for a particular product may depend on its price, customers’ incomes, the prices of substitutes, seasonality, marketing costs and the general economic environment. Production costs may depend on output volume, the price of energy, labour, exchange rates, wastage and capacity utilisation. A public institution’s budget revenue may depend on economic activity, employment, inflation, regulatory changes and the behaviour of taxpayers.

The model is not economic reality itself. It is a limited representation of selected aspects of that reality. Its value therefore depends on the correct formulation of the problem, the quality of the data, the validity of the assumptions, statistical identification, diagnostic checks and the robustness of the results beyond the period for which the model was built.

What is an economic forecast?

An economic forecast is a conditional estimate of a future value or trend, derived from available data, the model and the specified assumptions. It may relate to sales, demand, prices, costs, inflation, employment, revenue, capacity utilisation, cash flows or any other measurable indicator.

A forecast is not a promise, nor is it a certainty about the future. It should include a reference date, a forecast horizon, the assumptions used and an assessment of uncertainty. Where the method permits, in addition to a central forecast value, we present a forecast range, alternative scenarios and sensitivity to key factors. As the forecast horizon increases, uncertainty generally rises, and the likelihood that structural changes will undermine historical relationships becomes greater.

A short-term assessment of the present or the immediate future may utilise up-to-date, high-frequency indicators and is referred to as ‘nowcasting’. Medium-term forecasting supports budgeting, capacity planning and strategic planning. Long-term projections are more dependent on scenarios and structural assumptions and should not be presented as an exact extrapolation of the past.

What is strategy within the context of this service?

Strategy is a reasoned choice of action in relation to a specific objective, constraints, time horizon and acceptable risk. An econometric model can show how the outcome changes under different conditions, but it cannot, on its own, determine which course of action is morally, legally or organisationally correct. This requires decision-making criteria.

For a company, the criteria might be sustainable growth, profit margin, liquidity, return on investment, market share or risk mitigation. For a public institution, it may be public benefit, value for money, accessibility, distributional impact, legality and sustainability. The same forecast may lead to different strategies depending on different objectives and varying risk-bearing capacity.

Therefore, a strategy encompasses not only a preferred scenario, but also implementation conditions, resource constraints, key performance indicators, action thresholds, a contingency plan and rules for updating the strategy when actual outcomes deviate from the forecast.

What is the difference between a model, a forecast, a scenario and a strategy?

A model quantitatively describes the relationships between specific variables. A forecast uses these relationships and current information to assess future developments. A scenario sets out a coherent combination of possible conditions, such as higher inflation, weaker demand and more expensive financing. A strategy determines what action should be taken under these conditions and according to which criteria.

These four levels are not interchangeable. A well-calculated forecast without a rule of action may remain unused. A strategy without a verifiable forecast may be based on wishful thinking. A scenario without clearly defined assumptions is merely a story, whilst a model with no link to a real-world solution is a technical exercise.

What problems can the service solve?

Forecasting demand, sales and revenue

The model can assess seasonality, trends, price sensitivity, the effects of economic indicators, promotions, regulations or changes in customer behaviour. The results support budgeting, production, stock management, supply chain management and sales targets.

Forecasting costs, prices and production capacity

Dependencies between volume, energy, raw materials, labour, exchange rates, logistics, productivity and unit costs can be modelled. This enables a comparison of capacity decisions, pricing scenarios and break-even points.

Macroeconomic, budgetary and socio-economic forecasts

For public institutions and organisations, forecasts and scenarios can be developed for revenue, expenditure, employment, demographic pressures, demand for public services, regional development or other indicators, provided there is sufficient data and a clearly defined objective.

Impact and Causal Effect Assessment

When the question is whether a particular policy, investment, price change or organisational measure has caused an observed outcome, it is not sufficient to compare the ‘before’ period with the ‘after’ period. It is necessary to construct a plausible counterfactual scenario of what would likely have happened without the intervention. A causal conclusion is drawn only if the data and the research design allow it to be identified.

Early warning and response rules

The model can monitor leading indicators and set pre-defined thresholds at which risk increases or a change is required in the budget, capacity, stocks, funding or policy. The early warning system must also take into account the cost of false alarms, as an overly sensitive model may trigger unnecessary actions.

Comparison of strategies

Alternative strategies are analysed under common assumptions and using the same evaluation criteria. This allows for the comparison, for example, of rapid expansion, phased investment, a wait-and-see approach, price changes, diversification, capacity constraints or various options for public policy measures.

Where in the economy can the service be applied?

Econometric modelling is not limited to a specific sector, product or type of organisation. It can be applied wherever there is a clearly defined economic question, a measurable outcome, relevant data and a decision that depends on future dynamics or on the assessment of a particular effect. The subject matter may be an individual product, a firm, a market, a sector, a region, a public system or a national economy.

Manufacturing and industry

A manufacturing enterprise can model orders, volume, unit costs, productivity, energy intensity, scrap, capacity utilisation or required working capital. This applies to the food and drink industry, mechanical engineering, the chemical industry, pharmaceuticals, electronics, textiles, building materials, automotive components and other manufacturing activities. Strategies may involve comparing expansion, automation, relocation, contract manufacturing, changes to the product mix or postponing investment.

Retail and consumer markets

A retail chain, distributor or online shop can forecast sales by product, outlet, region and channel, measure seasonality and price sensitivity, and plan stock levels, promotions and logistics. The model can show how income, inflation, competitors’ prices, weather, public holidays or marketing activities are related to demand, without automatically presenting every observed correlation as a causal effect.

Services, tourism and property

A professional services provider can forecast workload, revenue, customer retention and required capacity. A hotel, tourism organisation or transport operator can model bookings, visitor numbers, seasonality, prices and occupancy. In the property sector, rents, demand, vacancy rates, construction activity and sensitivity to income, interest rates and demographic changes can be analysed, without the model being presented as a guarantee of future prices.

Energy, agriculture and logistics

In the energy sector, consumption, load, prices, production and sensitivity to weather and economic activity can be forecast. In agriculture, models can link yields, costs, climatic factors, prices and demand, whilst taking into account high levels of natural and market uncertainty. In logistics, freight flows, lead times, capacity, route costs and the risk of disruption can be assessed.

Finance and institutional investment decisions

A financial institution or professional investor may use macroeconomic, sectoral and firm-specific scenarios for internal planning, sensitivity analysis and risk management. The analysis may examine interest rate, inflation, exchange rate and credit factors, but does not constitute portfolio management, personalised investment advice, brokerage or a guarantee of returns.

Public administration and public systems

A public institution can model budget revenue and expenditure, employment, demographic changes, regional development or the demand for transport, health, education, administrative or social services. When evaluating a policy, the model can compare continuing with the current regime, introducing a targeted measure, and an alternative allocation of resources. Strategic assessment encompasses not only the overall effect, but also the cost, probability, time to outcome, distribution across groups and regions, and the risk of unforeseen consequences.

These applications are examples, not an exhaustive list. The crucial question is not ‘which sector you work in’, but ‘what needs to be explained or predicted, using what data, and for which actual decision’.

Who is the service intended for?

Public institutions

The service is suitable for ministries, agencies, local authorities, public enterprises and other institutional clients for budgetary and strategic planning, forecasting demand for public services, policy evaluation, regional and sectoral scenarios, and the development of monitoring and early-warning systems.

Companies and industry

For manufacturers, trading companies, technology firms and sector organisations, the models can support sales and demand forecasting, capacity and stock management, budgeting, the assessment of pricing and cost factors, the selection of investment strategies and the management of operational risk.

Institutional and professional investors and organisations

For financial institutions, funds, corporate investors, professional associations and other organisations, the service can provide macroeconomic, sector-specific and company-specific scenarios, sensitivity analysis and an analytical basis for internal decision-making. It does not constitute portfolio management, personalised investment advice, brokerage or a guarantee of returns.

The service is intended for organisations and professional clients. It is not offered as a retail service to individuals acting outside their commercial or professional capacity.

When might the service be required?

Econometric modelling is necessary when a decision depends on future dynamics, on the interaction between several measurable factors, or on distinguishing a real effect from a random coincidence. It is particularly useful for budgeting, capacity planning, policy evaluation, investment decisions, demand forecasting, developing a pricing strategy or creating an early warning system.

Modelling is not appropriate simply because there is a lot of data. If the question is unclear, the indicators do not measure the necessary concepts, the time period is too short, or the environment has changed irreversibly, a more complex model may provide greater confidence, but not greater knowledge. In such cases, the first professional conclusion may be that a reliable forecast within the required scope is not possible and that further data, a shorter time horizon or a different method is required.

What data is required?

To begin with, it is necessary to define the specific action or decision that the forecast is intended to support, the indicator to be explained or forecast, the time horizon, the frequency of the data, possible factors, the actual alternatives and the criteria for success. ‘We want a forecast’ is not a sufficient brief. We need to know what will be done differently depending on the outcome. 

Internal data may include sales, quantities, prices, discounts, orders, customers, products, costs, energy, labour, stock, deliveries, capacity, wastage, deadlines, cash flows, budgets and previous forecasts. For a public contracting authority, this may include administrative, budgetary, territorial, demographic and operational data, provided that there is a legal basis and the necessary safeguards are in place.

External data may include official statistics, national accounts, inflation, interest rates, exchange rates, employment, income, trade, energy prices, meteorological factors, regulations, technological indicators and other relevant sources. The use of more variables is not automatically an advantage. Only factors with a theoretical, empirical and practical basis are included.

For each variable used, the following must be known: its definition, unit of measurement, reference period, frequency of observation, source, revision rules, missing values and changes in measurement methodology.

There is no universal minimum number of observations. The required volume depends on the frequency, time horizon, seasonality, number of factors, stability of the process and type of model. One hundred observations following a structural change may be more useful than a thousand observations from a regime that no longer exists. Before accepting an assignment, we carry out an assessment of analytical feasibility.

What model do we use?

The method is selected according to the question, the data and how the result will be used. For time series, trend and seasonality models, regression models, autoregressive and vector models, state-space models, combinations of forecasts or other appropriate methods may be used. For data involving multiple units and periods, panel models may be used. When conducting impact assessments, quasi-experimental or other causal-analytical designs may be applied, provided their assumptions are met.

Machine learning may be useful where there are many factors, non-linear relationships or large volumes of data, but it is not automatically considered superior to an econometric model. If the task requires a causal explanation, transparency or regulatory traceability, a more interpretable model may be more appropriate. Where justified, we compare several models or use a combined forecast.

In a forecasting task, a more complex model is justified only if out-of-sample testing shows a significant improvement over a suitable baseline model. For an explanatory or causal task, the choice is determined by the identification strategy, the assumptions, the interpretability and the validity of the causal inference, rather than solely by predictive accuracy.

How do we work?

1. We define the decision and the measurable outcome

We clarify what decision lies ahead, what needs to be predicted or assessed, who will use the result, what the time horizon is, and what margin of error is practically acceptable. The correct model of an incorrectly framed question is of no value.

2. We check the data and its significance

We check the origin, completeness, frequency, comparability, revisions, missing values, outliers and changes in definitions. We distinguish between a real economic signal and a measurement artefact.

3. We formulate the assumptions and the baseline forecast

We determine which relationships have a theoretical basis, which factors are present at the time of the forecast, and what simple baseline forecast needs to be exceeded. This safeguards against a model that appears convincing merely because it explains the past.

4. We evaluate the model and carry out diagnostics

We check the statistical specification, stability, residuals, sensitivity, multicollinearity, endogeneity, structural breaks and other risks, depending on the method. When sufficient data is available, we separate the training period from the validation period.

5. We test the predictive ability

We use backtesting or another out-of-sample test and compare the error with a suitable baseline model. We do not assess quality solely on the basis of how well the model describes the data on which it was trained. A model that perfectly replicates the past but fails on new data is over-fitted and does not provide a reliable basis for action.

6. We develop forecasts, scenarios and sensitivity analyses

We present a central forecast, a range or scenarios depending on the method and data. We examine how the result changes with different values of key factors and which assumptions account for the greatest proportion of the risk.

7. Transforming the results into a strategy

We compare alternative courses of action against common criteria. We identify a recommended option, the conditions for its validity, critical thresholds, contingency measures and indicators for follow-up monitoring.

8. We document and establish rules for updating

We provide a description of the data, the method, the assumptions, the diagnostics, the limitations and the method of interpretation. We determine when the model should be updated, reassessed or decommissioned.

What knowledge does econometric analysis generate?

Epistemologically, an econometric model does not transform the unknown into certainty. It formulates statements whose validity can be tested against the data, the method and their predictive power. We therefore distinguish between several types of knowledge.

Descriptive knowledge establishes what has been observed: levels, distributions, trends, seasonality, volatility and structural changes. It relates to specific data and a specific period and does not automatically provide an explanation as to why the observed phenomena have occurred.

Correlational knowledge shows that changes in one indicator are statistically linked to changes in another. This relationship may be useful for forecasting, but does not in itself prove causality. A third factor, reverse causality, selection or a change in measurement may produce the same result.

Causal and counterfactual knowledge answers the question of what effect a specific intervention would have on the plausible course of events without it. It requires identification logic and assumptions that must be defensible. Where these conditions are not met, we do not present the association as a proven effect.

Predictive knowledge is a conditional statement about the future. It indicates what is probable given the available information, the specified model and the specific assumptions. The forecast must be capable of being compared with subsequent reality, and the error measured. This makes it verifiable, but not infallible.

The strategic and normative conclusion answers the question of what course of action is preferable. It does not follow solely from the statistical result, but from its combination with objectives, legal and resource constraints, the cost of error, the distribution of consequences and acceptable risk. Different rational strategies may follow from the same facts if the objectives differ.

In plain language, this means that we show what has been measured, what relationship has been established, whether causality can be inferred, what is likely to happen, how uncertain the estimate is, and for which objectives a particular course of action is justified.

How do we measure uncertainty?

Uncertainty may arise from random variability, incomplete or inaccurate data, unknown parameters, model choice, future values of external factors and structural changes. These sources should not be concealed within a single figure.

Where applicable, we use forecast intervals, distributions, scenarios, sensitivity analysis, simulations and model comparisons. We present not only the most likely outcome, but also the range of realistic deviations, the factors contributing most to the risk, and the events under which the model should be reviewed.

Accuracy is assessed using pre-selected error metrics and in relation to the practical cost of the deviation. In the case of stocks, underestimating and overestimating demand can have different costs. In the context of a public budget, an optimistic error may lead to a shortfall, whilst an overly conservative one may tie up useful resources. Therefore, ‘most accurate’ does not always mean ‘most useful for decision-making’.

Advantages of econometric modelling

The main advantage is that it transforms implicit expectations into verifiable relationships and assumptions. The model makes it possible to measure the impact of key factors, compare strategies under the same conditions, assess uncertainty and establish pre-defined response rules.

It facilitates more consistent budgeting, capacity and resource planning, the identification of key performance indicators, impact assessment and reporting to management, investors, lenders, auditors or public authorities. When well documented, the model makes the decision reproducible: another competent analyst can trace the data, logic and limitations of the model.

The greatest practical value lies not in the forecast always being accurate – which is impossible – but in the error being visible, observable and linked to a pre-defined action.

Limitations and potential shortcomings

The reliability of the model cannot exceed the reliability and relevance of the data. Missing observations, changes in definitions, selection bias, measurement errors, short time series and non-representative samples may limit or distort the result.

Historical relationships may cease to hold following a technological breakthrough, regulatory change, war, pandemic, shift in consumer behaviour or other structural shift. The model may overlook an important factor, confuse cause and effect, or be over-fitted to the past. A longer time horizon increases the risk of such deviations.

Developing and maintaining a reliable model requires time, expert judgement, client involvement and ongoing monitoring. Blindly trusting a numerical result can be more dangerous than having no model at all. We therefore do not provide guarantees regarding the accuracy of forecasts, profits, returns, budgetary outcomes or the success of a strategy, nor do we present statistical significance as sufficient evidence of practical significance.

Is econometric modelling mandatory?

As a general rule, econometric modelling is not a universal legal requirement for every business, investment or public decision. It may be required by a specific regulatory framework, methodology, funding procedure, public procurement process, contract, internal rules, investment mandate, budgetary process or a decision by a competent authority.

Even where it is not formally mandatory, modelling may be practically necessary if, without it, a budget, demand forecast, capacity, or the expected effect of a policy or strategy with material consequences cannot be reasonably justified.

The service does not constitute investment advice, portfolio management, investment intermediation or any other regulated investment service in relation to financial instruments.

What does the client receive?

The final deliverable is specified in the individual contract. It typically includes an executive summary, a formulated analytical question, a description and assessment of the data, a model specification, assumptions, diagnostic checks, appropriate validation, a forecast or assessment of the effect under investigation, a range of uncertainty, scenarios, sensitivity analysis, strategic alternatives, constraints and rules for updating.

Depending on the objective, the model may be provided in an agreed file format, as tables, visualisations, an interactive dashboard, a presentation to senior management, a technical appendix, a data glossary, a set of indicators and a procedure for follow-up monitoring.

Sectoral, market, financial and operational indicators may be used as input variables or assumptions in the model. This does not automatically constitute a standalone sectoral or market analysis, a detailed financial analysis, a value-added analysis or an economic security analysis. Where reliable modelling requires such a standalone analysis, this shall be specified separately within the scope of the project.

The provision of programme code, modification rights, licensed data or operational integration is expressly set out in the tender and the contract and is not automatically implied.

How long does it take?

The timeframe depends on the quality and availability of the data, the frequency and length of the time series, the number of indicators, the need to link data sources, the type of model, the forecast horizon, the number of scenarios, and the documentation and integration requirements.

A preliminary assessment of analytical feasibility and data can take between 10 and 15 working days. A focused model for a single key indicator with forecasting and scenarios typically requires between 20 and 35 working days. A complex project involving multiple interrelated indicators, panel data, causal analysis, several organisations or countries, a dashboard or integration may require between 40 and 70 working days or more.

These are guidelines, not automatically applicable deadlines. The specific timetable begins once the scope has been finalised, the quotation has been accepted and sufficient data has been provided. Delays, changes or incompleteness in the input data may necessitate a change to the timeframe and methodology.

When is the model complete?

A model is complete not when it has reached its highest level of mathematical complexity, but when it is suitable for the issue at hand; the data and assumptions have been documented; the diagnostic checks have been carried out; the prediction error has been measured against a baseline; the limitations are clear; and the result can be translated into a concrete solution and monitoring rules.

The model does not remain valid indefinitely. Following implementation, actual values must be compared with predicted values, errors analysed, and in the event of structural change, a deterioration in accuracy or a change in objective, the model must be updated or replaced.

Do not start by asking ‘which is the most up-to-date model?’. Start with the decision that needs to be made, the cost of error, and the data you actually have available.

Frequently Asked Questions

What is econometrics in plain language?

Econometrics uses economic logic, mathematics and statistics to measure relationships in real-world data. It can, for example, investigate how price, income, seasonality and the economic environment are related to demand, and whether these relationships are stable enough to allow for forecasting or impact assessment.

What is the difference between data analysis and an econometric model?

Data analysis can describe what has happened using tables, indicators and visualisations. An econometric model formalises the relationships between variables, assesses their uncertainty and can be used for forecasting, hypothesis testing or causal inference, provided the assumptions are met. Not every analysis requires an econometric model.

What is the difference between a forecast and a scenario?

A forecast provides a central or probabilistic estimate of future developments based on a specific model, data, error criteria and assumptions. A scenario sets out a hypothetical combination of factors – for example, a recession, higher interest rates and lower demand – and shows what the outcome would be under such conditions. A scenario is not necessarily a claim that this combination is the most likely.

What can be forecast?

A measurable indicator can be forecast if there is sufficient relevant and comparable data and a reasonably stable process. This could include sales, orders, demand, prices, costs, revenue, employment, inflation, capacity utilisation, stock levels, budgetary indicators or demand for a public service. Feasibility and accuracy are assessed on a case-by-case basis, rather than by the name of the indicator.

Can a forecast provide an exact figure?

A central value can be calculated, but presenting it without any indication of uncertainty is often misleading. A professional forecast specifies a range, scenarios or probabilities, where the method permits, as well as the conditions under which the result remains valid.

How far ahead can the model forecast?

The forecast horizon depends on the process, the frequency of data and the objective. Short-term forecasts typically use more up-to-date information and may be more accurate. For a long-term horizon, structural assumptions and scenarios play a greater role, and the range of uncertainty widens. We do not set a horizon longer than that which the data and the decision can reasonably support.

How much historical data is required?

There is no universal minimum. The required number of observations depends on the frequency, seasonality, the number of factors, structural changes and the type of model. A monthly forecast with seasonality usually requires more than one annual cycle, but the number of periods alone does not guarantee validity. An assessment of the data and analytical feasibility is carried out first.

Can the model be developed without internal data?

Sometimes yes, if the issue can be investigated using sufficient official, public or licensed sources. However, without internal data, the ability to forecast the organisation’s specific outcome is usually more limited. This is explicitly stated and may necessitate a more general sector-based model or a broader forecast range.

How do you choose the right model?

The choice is based on economic logic, the structure of the data, the forecasting horizon, the need for causal explanation, interpretability requirements and the results of validation. We compare suitable alternatives and a baseline. We do not assume that a newer or more complex algorithm is automatically better.

Can the model demonstrate a causal relationship?

Only if the data and research design allow the effect to be identified against a plausible counterfactual scenario. Simple correlation or good predictive ability are not sufficient evidence of causality. When a causal inference is not justified, the result is presented as a relationship or a predictive signal.

Do you use artificial intelligence and machine learning?

Yes, when they are appropriate for the specific question and data. They can assist with information processing, the discovery of non-linear relationships, model comparison and automation. Key decisions regarding data, methodology, validation and interpretation remain under the control of human experts. Artificial intelligence is not used as an unverifiable authority and bears no professional or legal responsibility.

How do you check whether a forecast is reliable?

Where the data allows, we set aside a period that the model has not seen, carry out backtesting and measure the forecast error. We compare the result with a simple baseline forecast and check its stability in the face of changes to the time period, factors and assumptions. Reliability is assessed for a specific horizon and objective, rather than as a general property of the model.

What happens in the event of structural change or a crisis?

In the event of structural change, historical relationships may weaken or disappear. In such cases, the forecast must be reviewed, the model reassessed, and the scenario logic updated. No model can predict all unknown shocks in advance. A good system identifies indicators that show when the old model is no longer reliable.

How does a forecast become a strategy?

The realistic alternatives, the objective, the constraints and the cost of various errors are identified. The results are then compared under common assumptions, and a preferred course of action, critical thresholds, a fallback option and monitoring indicators are formulated. The strategy is conditional on the objectives and is not automatically derived from the model.

Is the service suitable for public institutions?

Yes. It can support budgetary and macroeconomic forecasts, policy evaluation, demand for public services, regional scenarios, resource allocation and early warning. The scope must be tailored to the contracting authority’s powers, the applicable law, the quality of administrative data, and transparency and accountability requirements.

Is econometric modelling mandatory?

There is no general rule making it mandatory for every decision. It may be required by a specific legal framework, methodology, procedure, contract, public procurement contract, funding body or internal rule. Regardless of any formal requirement, in the case of a significant decision, the model may be practically necessary for verifying assumptions and assessing risk.

What exactly does the client receive?

The standard deliverables include an analytical report, a description of the data and methodology, a model, diagnostics, out-of-sample validation, a forecast, a range of uncertainty, scenarios, sensitivity analysis, strategic conclusions and updating rules. The specific files, visualisations, presentations, dashboards, code and usage rights are set out in the individual quotation and contract.

Can the model be integrated into Excel, a dashboard or an information system?

Yes, where this is technically and legally permissible and is included within the scope. The format depends on the complexity, update frequency, data sources, access requirements and the need for an audit trail. Operational integration, support and licences are negotiated separately from the analytical model itself.

How long does development take?

Data assessment and feasibility analysis typically take 10–15 working days. A focused model for a key indicator with forecasts and scenarios may take 20–35 working days. A complex project involving a causal analysis, multiple indicators, countries, organisations or integration may require 40–70 or more working days.

The timeframe indicated is approximate and begins once the necessary input data has been received. Where information is obtained from external sources or under the Access to Public Information Act, or where data is incomplete or contradictory, the timeframe may be extended.

How much does an econometric model cost?

The price is determined on a case-by-case basis depending on the nature of the enquiry, the quality and availability of the data, the number of indicators and scenarios, the methodological complexity, the forecast horizon, the need for licensed sources, visualisation, integration, documentation and support. Following the initial consultation, the client receives a quotation setting out the specific scope, deliverables, timetable and price.

How is confidential information protected?

We request only the data necessary for the agreed purpose, restrict access to it and apply the agreed rules on confidentiality, data protection and trade secrets. Sensitive or confidential information should not be sent via an unsecured channel until a suitable method of transmission has been agreed.

How often should the model be updated?

The frequency depends on the rate of change, the frequency of new data and the cost of error. An operational model may be updated monthly or quarterly, whilst a strategic model should be updated at the start of a new budget cycle, following a significant change in assumptions, or upon a predefined trigger. In addition to scheduled updates, there should be a threshold for a deterioration in the forecasting error that triggers a reassessment.

Does the model guarantee future results?

No. The model can improve the evidence base for a decision, but it cannot eliminate uncertainty, predict all shocks or guarantee sales, profit, returns, budgetary outcomes or policy success. The professional value lies in clear assumptions, measured error, the range of risk and timely updates.

How does the commissioning process begin?

The first step is a brief description of the analytical question, the indicator to be explained or forecast, the forthcoming choice or action that depends on the result, the time horizon, the frequency and availability of data, and the deadline. Following an initial assessment, SOLISFERADOME ANALYTICS determines the analytical feasibility, the data required, the appropriate method, possible scenarios, the expected outcome, the timeline and the individual price.