Core models (frequently used)

  • External sector models

  • Econometric models for the Chilean economy

  • Structural models for the Chilean economy

  • Estimating unobservables

External sector models

The first stage in the preparation of the central projection scenario involves the construction of the international scenario, which includes the forecasts of the international macroeconomic variables that are most relevant for the Chilean economy. These forecasts feed the models used in the short- and medium-term projections of the national economy. The variables projected within the international scenario are divided into three blocks:

Real activity and external prices:

  • GDP growth in the main economies and regions of the world
  • Inflation of Chile’s main trading partners

Commodities

  • Copper prices
  • Fuel prices
  • Food prices

Financial markets:

  • Projected U.S. 10-year bond rate and monetary policy rate.
  • Projected three-month Libor rate in US dollars.
  • Expected volatility index (VIX)
  • Latin American sovereign bond risk premium index (EMBI).
  • U.S. multilateral real exchange rate.

Econometric models for the Chilean economy

The central scenario of short-term forecasts is constructed by combining multiple estimates generated from econometric models, which are supported by external conjunctural information and the staff's judgment.

The Central Bank uses different methodologies to forecast aggregate GDP and its components, employing expenditure and/or supply-side approaches. Separating GDP into its components has two advantages. On the one hand, it allows for a more detailed analysis of the particular dynamics of the different economic sectors. On the other, the aggregation of the forecasts for the different components makes it possible to obtain an alternative forecast for total GDP.

In an inflation-targeting scheme, it is essential to forecast the evolution of prices, which includes the evaluation of total CPI and its different components, the core inflation component being of special importance, given its greater stability and high correlation with medium-term inflation.

The macroeconomic consistency model (MACRO) is a set of macroeconomic identities that underlie the national accounts accounting approach. It ensures consistency of macroeconomic forecasts with the basic macroeconomic identities that characterize the national accounts. The MACRO model forecasts include real aggregate demand and nominal aggregate demand, together with their components, export and import flows, among others.

Structural models for the Chilean economy

At the Central Bank, structural models play basically two roles: they generate the central forecast scenario for the medium term, and they are instrumental in the analysis of alternative scenarios and in the evaluation of monetary policy consistent with such scenarios.

The Central Bank has two medium-term models for the local economy: i) the semi-structural projection model (MSEP), and ii) the extended structural model for analysis and simulation (XMAS). Both approaches seek to contribute to a better understanding of the forces guiding the expected evolution of the economy, considering interrelationships between key economic variables, such as output, inflation and interest rates. The complementary use of models with different characteristics, such as XMAS and MSEP, is based on the search for greater robustness in the analysis and projections, providing different viewpoints that, complemented by the judgment of the Board and the economic staff, help to generate a global vision of the state and future prospects of the economy.

In general, structural models such as XMAS are anchored on parametric constraints and equations based on economic theory fundamentals, making them more robust to Lucas critique, and allowing for robust evaluations of policy and counterfactual scenarios.
In terms of scale, XMAS has greater complexity in its structure, enabling a more detailed analysis of different shocks and propagation channels.

XMAS is a structural model that is an extension of the MAS model developed by Medina and Soto (2007), and formerly used in the Central Bank's analyses and forecasts. Unlike other empirical models of a purely econometric nature, structural models such as XMAS are based on the explicit modeling of the behavior of different agents in the economy, relying on the fundamentals of economic theory regarding the decisions of households and firms. This allows not only to generate a likely trajectory for the future of the economy, but also to give a more complete interpretation of the forces behind the projections.

Semi-structural models such as MSEP place greater emphasis on the econometric fit to the data, at the expense of relaxing some theoretical restrictions and greater vulnerability to Lucas critique. It is a simpler model, which privileges parsimony and flexibility over complexity, making it easier to follow the modeled transmission mechanisms.

The MSEP, described in detail in Arroyo et al. (2020), represents an advance with respect to the MEP model documented in Central Bank of Chile (2003) and in García et al. (2005). It is a simple stylized model, adapted to the Chilean reality, which includes different econometrically estimated relationships and is framed within gap models. These models explain the cyclical dynamics of an economy, that is, when certain variables stand above or below their trend values, thus exhibiting positive or negative "gaps." As it is a model written in gaps, the trends of the variables are calculated at a stage prior to the use of the model, and the relationships between the different variables are restricted to their cyclical components.

Estimating unobservables

An important part of monetary policy decisions are based on gaps between observable variables, such as the interest rate or GDP, and some unobservable counterpart, such as the neutral interest rate and potential and/or trend GDP, respectively. Thus, for example, the interest rate will be contractionary, helping to contain inflation, if it is above its neutral level. Similarly, activity will be in an expansionary cycle, generating inflationary pressures, if actual GDP is above potential.

The correct estimation of these variables is of utmost importance to correctly evaluate monetary policy. These variables cannot be observed directly and must be inferred by a combination of statistical methods and economic theory. This inference obviously has a significant degree of uncertainty, as it is based on estimates from multiple models, each with its own degree of uncertainty. The judgment of the staff and the Board determines, through a critical analysis of the different estimates, the values to be used to calibrate monetary policy. The estimates are updated annually, and the results are made public in the monetary policy reports. For details, see “Use of Macroeconomic Models at the Central Bank of Chile.