Approaches to Economic Forecasting

While the long-run trend growth rate (covered in our TFP article) reflects the supply side of the economy, most macroeconomic forecasting focuses on short- to intermediate-term fluctuations around the trend — the business cycle. These fluctuations are mostly driven by shifts in aggregate demand, with short-term aggregate supply shifts playing a smaller role.

The CFA Level III curriculum identifies three distinct forecasting approaches:

Loading diagram...

These approaches are not mutually exclusive. Thorough analysis typically incorporates elements of all three.

Approach 1 — Econometric models

Econometrics is the application of statistical methods to model relationships among economic variables. The two main flavors:

Type	Theoretical grounding	Example
Structural	Functional relationships derived from economic theory	A New Keynesian DSGE model with explicit consumption and investment equations
Reduced-form	Compact representation of underlying structural model, or purely data-driven	A vector autoregression (VAR) of GDP, inflation, interest rates

Econometric models range from small (a handful of equations) to large (hundreds of equations). All work the same way: the analyst supplies values for exogenous variables (e.g., exchange rates, commodity prices, policy rates) and the model produces forecasts for the endogenous variables.

Loading diagram...

Strengths

• Discipline and consistency — the model imposes internal coherence
• Quantitative simulation — change one input, see the systematic effect
• Challenges priors — the model output may contradict the analyst intuition, forcing reassessment
• Fast updating — once specified, new data flows through automatically

Weaknesses

• Complex and time-consuming to specify initially
• Input forecasts are themselves uncertain — error in exogenous variables propagates
• Mis-specification risk — relationships change over time, the assumed structure may be wrong
• False precision — the model gives numeric output that can hide deep uncertainty
• Poor at turning points — econometric models rarely call business cycle peaks and troughs

The pragmatic adjustment

When forecasters notice systematic forecast errors, the disciplined approach is to overhaul the model. The pragmatic alternative — and what most practitioners actually do — is to incorporate past forecast errors as an additional explanatory variable. This is a tacit admission of mis-specification, but it can improve accuracy in the short run.

Approach 2 — Economic indicators

Economic indicators are statistics that contain information about the economy past, present, or future position in the business cycle.

Loading diagram...

Most analysts focus on leading indicators because they purport to predict future activity, inflation, interest rates, and security prices.

The OECD composite LEI

The Organization for Economic Co-operation and Development (OECD) publishes a composite leading indicator for each country or region, built from 5 to 9 variables such as:

• Share prices
• Manufacturing metrics (new orders, hours worked)
• Inflation
• Interest rates (slope of yield curve)
• Monetary aggregates

These variables exhibit cyclical fluctuations similar to GDP, with peaks and troughs occurring 6 to 9 months earlier with reasonable consistency.

The diffusion index

Individual LEIs can be combined into a diffusion index that measures how many indicators are pointing up vs. down:

$$\text{Diffusion} = \frac{\text{Number of indicators pointing up}}{\text{Total number of indicators}}$$

If 7 out of 10 indicators are pointing upward, the diffusion index is 70% — the odds favor an accelerating economy. Below 50% suggests deceleration.

Look-ahead bias — the biggest weakness

When the OECD revises the LEI methodology each month, the entire historical series is restated. As a result, the most recently published historical series will appear to have fit past business cycles more accurately than it actually did in real time. This is look-ahead bias — the LEI looks better in backtests than it performs in live forecasting.

Nowcasting

After the 2008 global financial crisis, a new methodology called nowcasting emerged. The best-known example is the Federal Reserve Bank of Atlanta GDPNow, launched May 2014. The goal: forecast the current quarter GDP based on data released throughout the quarter, BEFORE the official BEA estimate.

Loading diagram...

BEA GDP release sequence

Estimate	Released	Notes
Advance	4 weeks after quarter end	Greatest market impact; what GDPNow targets
Preliminary	~1 month later	Revised with more data
Final	End of following quarter	Most accurate but stale

Nowcasting strengths and limitations

Strengths: real-time updating, focused on a single variable of primary interest (GDP).

Limitations:

• Not predictive beyond the current quarter
• Highly volatile early in the quarter (few data points)
• By the time it stabilizes, it has lost much of its forecasting edge

Approach 3 — Checklist

The checklist approach involves continually monitoring a wide range of economic data and assessing whether each measure is in equilibrium or at an extreme. Data may be extrapolated via statistical methods (e.g., time series) or judgmentally.

Loading diagram...

Strengths

• Flexibility — easy to add, drop, or reweight variables
• Structural-change tolerance — can adapt quickly when relationships break down
• Breadth — can include any topic, perspective, theory

Weaknesses

• Subjective and arbitrary — no formal mechanism for combining data
• Time-consuming — manual process
• Inconsistent — different items may be weighted differently at different times
• Cognitive bias vulnerable — what looks "interesting" to the analyst may not be what matters

Comparison: strengths and weaknesses

Approach	Strengths	Weaknesses
Econometric models	Discipline, consistency, quantitative simulation, scalable	Complex, mis-specification risk, false precision, weak at turning points
Leading indicators	Simple, intuitive, focused on turning points, third-party availability	Look-ahead bias, frequent revision, false signals, binary directional guidance
Checklist	Flexible, accommodates structural change, broad	Subjective, time-consuming, inconsistent, biased

The Izek vs Berke example

The curriculum gives a worked illustration of two analysts at Cycle Point Advisors:

Loading diagram...

Berke approach

• Time series model uses a published LEI series as a key input
• Presents econometric forecasts at each meeting
• Approach: econometric model + LEI hybrid

What attracts him: quantitative output, consistency, discipline.

Weaknesses he may overlook:

• Model mis-specification — could produce systematic forecast bias
• False precision — even an unbiased model has wide forecast errors
• Look-ahead bias in the LEI input — the historical series appears more accurate than it was in real time

Izek approach

• Samples a wide variety of research monthly
• Focuses on whatever perspectives seem most interesting that month
• Brings a stack of charts to each meeting
• Approach: essentially a checklist

What attracts her: flexibility — can include anything she finds interesting.

Weaknesses she may overlook:

• Subjectivity and judgment — no clear mechanism for combining signals
• Idiosyncrasy — her "checklist" depends on what is salient to her
• Cognitive bias — basing the checklist on what is "most interesting" in others' research makes her process vulnerable to recency, availability, and herding biases

The CFA-exam pattern

Module 1.05 questions typically ask you to:

1. Identify the approach given a description of an analyst process
2. Match approach to strength — when to prefer econometric vs. LEI vs. checklist
3. Identify weaknesses — what could go wrong with the chosen approach
4. Recognize biases — look-ahead bias in LEIs, mis-specification in models, subjectivity in checklists
5. Combine approaches — recognize that thorough analysis uses elements of all three

Recall that the three approaches are complementary, not substitutes. The best forecasters use:

• An econometric model for internal consistency and quantitative simulation
• Leading indicators for turning-point detection
• A checklist for structural-change adaptation and breadth

For the upstream growth-accounting framework that defines the trend around which the cycle fluctuates, see our TFP article. For the CME application, see our growth-application article. Practice forecasting-approach questions in our CFA Level III question bank.