The decline of home cooked food ‐

The share of home-cooked food in the diet of UK households declined from the 1980s. This was contemporaneous with a decline in the market price of ingredients for home cooking relative to ready-to-eat foods. We consider a simple model of food consumption and time use that captures the key driving forces behind these apparently conﬂicting trends. We show that observed behaviour can be rationalised by the fact that the shadow price of home-cooked food, which accounts for the fact that cooking takes time, has risen relative to the price of ready-to-eat food, due to the increase in the market value of time of secondary earners. We discuss the implications for policies that aim to encourage healthier diets.


INTRODUCTION
Households in the developed world have shifted away from home-cooked food towards ready-to-eat food. In the UK, home-cooked food represented more than half of the food budget in 1980, but less than a third in 2000. This is an enormous change, which has been associated with equally important changes in life-style and time use, and in particular with changes in labour market behaviour of secondary earners. This change in diet has been proposed as one of the likely candidate causes of the growth in obesity and other non-communicable disease, 1 and there is considerable interest in understanding what has driven this change in order to help design policies to reverse it. Our contribution in this paper is twofold. First, we document a number of interesting trends. Both the price of ingredients for home cooking and that of ready-to-eat alternatives (at home) decreased over the period 1980-2000, with ingredients decreasing by relatively more. Assuming that both types of food are normal goods, then we would expect the consumption of ingredients to increase, both because of the substitution and the income effect. Real income growth in the period would reinforce these effects. However, we show that real consumption of ingredients has fallen.
Our second contribution is to show that an increase in the opportunity cost of time for secondary earners (largely females) helps to explain the shift from ingredients used for home cooking to readyto-eat foods. Using the insight from Becker (1965), that consumption comes from the combination of market goods and time, a simple model of food demand and time use with home production allows us to recover the shadow price of home-cooked food. This incorporates the opportunity cost of time as well as the price of ingredients and returns to scale in food preparation. We estimate the opportunity cost of time of secondary earners and show that the rising opportunity cost of time has increased the shadow price of home cooking, which helps to explain the shift from ingredients used for home cooking to ready-to-eat foods.
The decline in home-cooked food is contemporaneous with a rise in labour market participation and wages. We show that in the cross-section, the expenditure share on home-cooked food is negatively correlated with female employment and wages, and that time spent on food preparation is correlated with household characteristics. These correlations point to the possibility that improved labour market opportunities for secondary earners and a reduction in household size led to a reduction in demand for home-cooked food.
In order to better understand the implications of these trends, we write down a simple model of demand that incorporates home production of food with heterogeneous time costs and two adults contributing time to home cooking. We incorporate the trade-offs between purchasing ingredients for cooking or purchasing ready-to-eat food. The model allows us to recover the shadow price of homecooked food, which incorporates the opportunity cost of time as well as the price of ingredients and returns to scale in food preparation.
We use the UK Family Expenditure Survey (FES) and distinguish labour-intensive and non-labourintensive food. We are able to track consumption, prices, wages and labour market participation over the period 1980-2000. Women's employment rate and hours worked have increased, as have real wages, making time spent cooking more costly in terms of foregone earnings. To find the shadow price, we have to impute the value of time (i.e. a potential wage for the individuals who are not working). We estimate a Heckman selection equation to recover wages for women not participating in the labour market. Putting this together with market prices, we show that increases in the opportunity cost of time help to explain the shift from ingredients used for home cooking to ready-to-eat foods, because the rising opportunity cost of time has increased the shadow price of home cooking.
Our paper relates closely to the literature that establishes the importance of non-separabilities between consumption and time use. 2 For example, Blow, Lechene and Levell (2014) reject separability between time use and consumption in preferences using data for the US. Our model is in the tradition of Barten (1964), in specifying that household composition acts as a price deflator. We follow Deaton and Paxson (1998), with economies of scale in food consumption, and Crossley and Lu (2018), with economies of scale in food preparation.
Our work is motivated by the literature that shows that the increase in the consumption of readyto-eat or processed foods has been linked to adverse health outcomes, such as obesity, as well as to negative impacts on cognitive outcomes, particularly amongst children. 3 This has led to calls for policy intervention aimed at changing eating habits, 4 and promoting a healthy, balanced diet -most recently 2 Browning and Meghir, 1991. 3 See Adams et al. (2020), Case, Lubotsky and Paxson (2002), Heckman (2007), Levine (2003a, 2003b), Baum andChou (2011), Cawley (2000), Goldman, Lakdawalla and Zheng (2009), Herbst and Tekin (2011) and Mackenbach et al. (2008). in the UK by the establishment of the independent National Food Strategy. 5 It is also motivated by the literature that seeks to understand why obesity has increased, and what role prices have played. 6 The structure of the paper is as follows. In Section 2, we describe trends in the evolution of food consumption and market prices. In Section 3, we summarise changes in labour market participation, wages and time use. In Section 4, we present a simple model of food consumption and time use with home production and discuss the implications for shadow prices. A final section provides a discussion of the implications.

TRENDS IN FOOD CONSUMPTION AND MARKET PRICES
We use data on expenditure, wages, employment and hours of work for a sample of 27,193 households from the UK FES. Details about these data are provided in online Appendix A.1. We restrict our attention to households with two adults and any number of dependent children (including zero), where both adults are of working age (25-60), and where the head of household works full-time. We do this because selecting on households where the head works full-time allows us to treat the hours of the main earner as exogenous, and simplifies modelling considerably. Employment rate of male heads of household in the age range 25-60 in the UK is high, and there is very little variation in hours worked (conditional on working full-time), meaning that the assumption that hours are exogenous is not very restrictive. 7 Importantly, our analysis does not include single-adult households. It is possible that time spent cooking varies for this group; for example, single-adult households might cook less due to lack of scale. An increase in the share of single-adult households could have contributed to the rise of processed food in aggregate, and could have important implications for policy; this is not something we study here.
We use data for the period 1980-2000; we stop in 2000 for a number of reasons. First, the main change in labour market participation that we are interested in occurred over the 1980s and 1990s, and the shares of ingredients and processed food are relatively stable after 2000. Second, the way the data were collected changes in 2000, making the time series of disaggregated products awkward to compare over this period.
In order to empirically investigate the shift away from home cooking, we have to be able to distinguish ingredients for home cooking from processed foods. The distinction that we make in this paper is in terms of time required to prepare food before it can be eaten versus foods that can be eaten with minimal preparation time. To our knowledge, this distinction has not been implemented with detailed household-level data, and one of the contributions of this paper is to make this empirical distinction. However, the distinction is not a clean one, as some foods can both be eaten raw and combined with time to make meals. Milk is an obvious example. For our purposes in this paper, we have taken the view that a broad categorisation into three categories is sufficient. For other purposes, for example, for estimating a structural model of demand, it is likely that further disaggregation would be necessary.
The FES data record details of expenditure on 367 food categories. We map these into three categories: ingredients, which typically can be eaten as they are and do not need to be combined with time to make home-cooked food; processed foods that are eaten at home, which either can be F I G U R E 1 Real food expenditure shares by food category Note: Sample of 27,193 households with two adults aged 25-60, with any number (including zero) of dependent children. Lines are fitted local polynomials. Shares are expenditure shares in constant 1980 prices. eaten in their natural state, or which have already been combined with time (by a firm) to make them edible, and so take minimal time input; and food out, which has already been combined with time (by the retailer) to make it edible and is eaten outside of the home (meals out, takeaways and snacks). This categorisation is summarised in Table A.2 in online Appendix A.
We use the retail price index (RPI) price series of the Office for National Statistics (ONS) to construct a price index for each of the food categories, as indicated in Table A.2. The price data consist of about 30 series of price indices obtained from the ONS. 8 The share of real expenditure on ingredients for home-cooked food declined dramatically from 1980 to 2000 (see Figure 1). In 1980, on average, 57 per cent of the food budget was allocated to ingredients for home-cooked food, with the remainder split between food purchased ready to eat at home and meals out, takeaways and snacks. By 2000, the share of ingredients for cooking at home had fallen to 35 per cent. The share of processed food for home consumption had risen from 26 per cent to 45 per cent, while the share of expenditure on meals out, takeaways and snacks has remained fairly stable at just under 20 per cent.
Market prices for ingredients for home cooking and processed foods for home consumption fell over the 1980s and 1990s (relative to the price of non-food items). The market prices of ingredients fell by around 30 per cent, and the prices of pre-prepared foods by around 20 per cent. Over the same period, the price of food out increased by close to 20 per cent (more for takeaways and snacks than for meals out), as shown in Figure 2.
In real terms, the average expenditure on ingredients shrank by 19 per cent over the period, while that of processed food increased by 89 per cent (see Figure A.1 in online Appendix A). This leads to a puzzle. The prices of both ingredients and pre-prepared food declined from 1980 to 2000, yet the consumption of ingredients has fallen while that of pre-prepared food has increased. It is not possible to rationalise these trends with standard models of demand, where food is typically either modelled as one composite good in preferences, or home-cooked food and processed food are substitutes in preferences. The relative decline in price should cause a substitution from pre-prepared food towards ingredients. Assuming that both are normal goods, then income growth over the period should also lead F I G U R E 2 Change in market prices of foods Note: Lines are the expenditure-weighted aggregated price indices for the relevant categories shown in Table A.2. The outside good includes all non-durable non-food consumption, including alcohol, tobacco, household services, personal goods and services, and leisure goods and services.
to increases in the real quantities consumed. 9 Thus, the decline in the real consumption of ingredients is particularly puzzling.
One important difference in these goods is that ingredients require time for preparation, while processed foods are pre-prepared and require no time (or at least much less time than ingredients). Therefore, changes in the opportunity cost of time are an obvious candidate to help explain these trends. Another possibility is that the quality and variety of processed food may have improved a lot more than ingredients. This could be interpreted either as the observed price increase in processed food overstates the truth, or as a reason for preferences to shift towards processed food. If the true qualityadjusted price of processed food has fallen by more than 10 per cent below the official measure, then, in theory, the substitution effect could lead to a shift from ingredients to processed food. However, it would take a much bigger measurement error and a large substitution elasticity to cause the huge shift we observe in Figure 1. Meanwhile, we do not have a strong reason or evidence base to think the ONS's own attempts at quality adjustment in constructing their price indices are severely inadequate. Similarly, it is hard to find evidence for how preferences have shifted. One could even make the opposite conjecture that preferences have shifted towards home-cooked food as consumers become more health conscious. Therefore, we take the neutral stance of assuming stable preferences and using official measures of prices.

TRENDS IN EMPLOYMENT, WAGES AND TIME USE
The 1980s and 1990s saw substantial changes in women's uses of time. Across all females, the employment rate was relatively low to start with (57 per cent in 1975) and increased substantially to close to 75 per cent around the mid-and late-2000s. 10 In our sample, the female employment rate was around 55 per cent, but this has increased significantly over this period to about 80 per cent in 2000 (see Figure 3(a)). Conditional on working, average weekly hours for women have increased from about 22 to 33 hours (see Figure 3(b)). Altonji and Blank (1999) and Costa (2000) document similar F I G U R E 3 Labour market participation of females Note: The dots in panel (a) show the proportion of women who were working each month, and the line is a local polynomial fit. In panel (b), the dots are the weekly working hours of women conditional on working, averaged for each month, and the line is a local polynomial fit.
trends in female labour market participation for the US. Most working-age males in the UK worked in the labour market over this period, and most worked full-time.
Real wages have grown for both males and females, but more so for females, as shown in Figure 4. Wage growth over this period has been around 40 per cent for females and 20 per cent for males.  From the TUS, we have a measure of time spent on food management by a sample of 1,005 males and females living in couples, where both adults are aged 25-60 and the man is employed full-time (excluding self-employed). From the PAUT, we have a sample of 408 couples, where both adults are aged 25-60 and the man works full-time.

Time use
In 2000, females spent on average 8.3 hours a week on food management as the main activity, while the male average (including zeros) is 3.3 hours (see Table A.3). The gender gap is much bigger in the subsample of couples where the female does not work, than in the subsample where the female works.
In 1974-75, the average time spent by females on food management was higher at 13.3 hours (compared to 8.3 hours in 2000). In online Appendix A.2, we show that this decline in female hours spent on food preparation is observed across the distribution, not only at the mean. By contrast, the average hours that males spent on food management increased from 1.3 to 3.3 over the period, a big proportional increase, but there remains an overall decline in the total hours spent by the couple on food management.
These trends are also observed in the US. Bianchi et al. (2000) document a 12.5 hours per week reduction in total female housework hours between 1965 and 1995. About two-thirds of that overall reduction comes from cooking meals and meal clean-up (8.5 hours). Similarly, Smith, Ng and Popkin (2013) documented that between 1965-66 and 2007-08, the amount of time spent in food preparation more than halved for females and nearly doubled for males in the US.

3.2
The correlation between home cooking and household characteristics Table 1 shows that women in work spend less time cooking on average than women not in work. Therefore, the increasing share of women in work would, in itself, reduce the average level of consumption of ingredients. We also see that real consumption of ingredients has fallen for both groups, but is lower and has fallen by more for households where the female works (see Figure 5). We expect families with a lower cost of time and a larger number of children to choose more homecooked food. This intuition is confirmed by cross-sectional correlations in both expenditure data and time use data.
In Table 2, we report conditional correlations between the share of ingredients in food consumption and household characteristics. We see that the share of ingredients (home-cooked food) is positively

F I G U R E 5 Real household consumption of ingredients, by female employment
Note: Local polynomial fit of real consumption of ingredients, in January 1980 prices. correlated with the number of children, negatively correlated with female employment, her hours (if working), and both male and female wages (proxies for the cost of time). Cross-sectional correlations in the 2000 TUS paint the same picture; details are provided in online Appendix A.2. Conditional on basic demographics, women in work spend about three fewer hours per week on food management. Among two-earner couples, an additional hour that the woman is at work is associated with a 0.1 hour reduction in her time on food management. Women's time spent on food management is negatively correlated with their own wage, and positively correlated with the number of children. Interestingly, men's time on food management is not strongly correlated with the number of children.

A SIMPLE MODEL
In order to investigate these correlations further, we write down a simple model that allows us to recover the shadow price of cooking and to see how it has changed over time.
We model consumption and time use, with home production of food. Households consist of n people, equal to two adults with any number K of children (including none). Utility is derived from household food consumption f (which is private); a non-food non-durable composite good x, which exhibits some degree of publicness; and leisure l. We allow for heterogeneity by allowing demographic composition to enter as price deflators following Barten (1964), Deaton and Paxson (1998), Crossley and Lu (2018) and Lewbel and Pendakur (2017). We specify a utility function that corresponds to a unitary model of the household: Here, ∈ [0, 1] captures potential returns to scale in the non-food good x; if = 0, then x is entirely public, and if = 1, then there are no returns to scale in x. Leisure enters preferences as the sum of the leisure times of both adult household members, l = l 1 + l 2 . Leisure times are assumed to be perfect substitutes in preferences. We assume that male working hours are constrained, so that the opportunity cost of leisure time for men is the wage of the women, and non-working time is the sum of leisure time and cooking time. Food can be cooked at home, by combining time and market-bought ingredients, or purchased ready to eat, in which case it requires no processing time. Home-cooked food and ready-to-eat food are not assumed to be perfect substitutes in preferences: where c is home-cooked food and r is ready-to-eat food, both at the household level. We follow Hamermesh (2008) in assuming that ingredients i and time spent cooking t are complements, so that home-cooked food c is produced according to We assume that the production technology is linear homogeneous in time and ingredients, but not in household size, so as to capture the fact that a home-cooked meal for two takes less than twice the time required to prepare a meal for one. In other words, there are returns to scale in cooking, which are represented by ∈ [0, 1]. If = 0, then it takes the same time to cook a given quantity of food per capita, whatever the number of people catered for, while if = 1, there are no returns to scale in cooking, so that it takes twice the time to cook for two as it takes to cook for one. The time inputs of the adults are perfect substitutes in the production of home-cooked food, t = t 1 + t 2 . The parameter B transforms quantities into time.
Adults allocate time between market work h s , the production of home-cooked food t s and leisure l s , with s = 1, 2 for the adult members of the household. The time constraints for both individuals are Working hours for the main earner are assumed to be constrained: This assumption is justified by empirical evidence. Indeed, the elasticity of hours of work of males is low, which is usually interpreted as due to a constraint on male hours. Non-market time is the time not spent working for a wage -it is the sum of the time spent cooking and of leisure. Leisure is all the time that is not spent sleeping, cooking or working for a wage. Because food can be produced at home, by combining time and ingredients, there is no separability between food and time, or between food and other non-durable goods. Households purchase ingredients i, ready-to-eat food r and non-food x, which they fund with market work and non-labour income: Here, p k is the market price of good k, y 0 is unearned income, and w s , s = 1, 2 is hourly wage for the main and secondary earners. Households choose how much ready-to-eat food and home-cooked food to eat, how to use time, and how much to spend on the non-food good. Prices and wages are assumed to be exogenously determined. From the production function, we obtain the relationships between home-cooked food c and ingredients i, and between home-cooked food c and time spent cooking t. The Leontieff assumption yields We can substitute for ingredients i and time spent cooking t in the budget constraint. Because of the assumption that the time inputs of both household members are perfect substitutes in the production of home-cooked food, there is one price for the time input t. The relevant price for the time input t is the opportunity cost of the time of the household member who is not constrained on the labour market; it is w 2 , the wage of the woman, or secondary earner. The budget constraint in terms of full-time income, and final consumption, where the time of the primary earner is valued at the wage of the secondary earner, is therefore Let k * and p * k denote the quantity demanded for final good consumed k and its shadow price, respectively. The household's problem can be rewritten, in terms of quantities of final goods consumed and shadow prices, as where the shadow prices are related to the market prices as Our object of interest, the shadow price of home-cooked food, is Home-cooked food results from the combination of ingredients and time, so the shadow price of home-cooked food involves the market price of ingredients and the opportunity cost of time for the household, the wage of the unconstrained individual (because of the constraint on hours worked by agent 1, the non-market time of agent 1 is valued at the wage of agent 2). The shadow price of home-cooked food is increasing in the price of ingredients, in the wage of the unconstrained individual and decreasing in household size. It does not depend on the woman's hours worked; instead, the latter is an endogenous outcome of the model and is likely to be correlated with wages and prices. 13 The model does not yield any prediction on how leisure or cooking time is split within the household. The model only has implications for t 1 + t 2 , l 1 + l 2 , and because l 1 + t 1 is constrained by the man's work, any marginal change in t 1 is compatible with the same change in l 2 and the opposite change in l 1 , t 2 . We have chosen to abstract from intra-household allocation of time because our main motivation is to the explain the decline in home-cooked food consumption, regardless of who does the cooking within the household.
An increase in the wage of the secondary earner corresponds to an increase in the shadow price of home-cooked food. This leads to a decrease in the demand for home-cooked food, and an increase in the demand for ready-to-eat food, as per the substitution effect. The income effect goes in the same direction as the substitution effect for home-cooked food, and in the opposite direction for ready-toeat food.
As well as increasing in the opportunity cost of time, the shadow price of home-cooked food is decreasing in the number of people in the household. If cooking is more efficient in larger households, then the shadow price of home-cooked food will be lower for larger households. While there has been a reduction in household size in the UK overall, this is largely due to an increase in singleperson households. Because we analyse the behaviour of households composed of two adults and any number of children, economies of scale did not play an important role quantitatively for these types of households.
The shadow price of home-cooked foods is also reducing in cooking technology. We hold this constant as we do not have any information to suggest that this has changed substantially over this period.

Wages
One challenge we face in calculating the shadow price of ingredients is that we do not observe wages for secondary earners who are not currently in the labour market. Which secondary earners choose to participate in the labour market and which do not may be endogenous (i.e. related to demand or preference shocks). To get around this problem, we use data on all households to estimate a wage equation controlling for selection by estimating participation status. This allows us to compute wages for all secondary earners. Details of how we do this are provided in online Appendix B, and we provide a brief description of our approach here.
We estimate wages using the Heckman two-step estimation. Each individual has a potential wage, W p i , if they participate in the labour market, and a reservation wage, W r i , that dictates whether they participate. Variables that shift both participation and reservation wages include the woman's age, age squared, age cubic, the woman's education in seven bands and year dummies. Variables that shift only the participation wage include dummies for 11 regions of the UK based on where the household resides and interactions between year dummies and four broader region dummies, so as to control for local labour market conditions. Variables that shift only the reservation wage include a dummy for whether there are children present in the household, the number of children, a fifth-order polynomial of the age of the youngest child, the household's unearned income, income from benefits, and the man's wage, age, education, hours of work and occupation, market prices of the goods in our model, housing tenure, and interaction between year dummies and the presence of children.
Our estimated selection equation results are as expected: the probability to participate is increasing in the education of the secondary earner, decreasing in unearned incomes, and decreasing in the wages and hours of work of males. It is also lower for females who have children and increasing in the age of the youngest child. The potential wage is increasing in the education of the secondary earner and the age of the secondary earner until about 50 years of age, and decreasing afterwards.
We use the estimated coefficients from the wage equation, along with estimates of the scale parameter, to predict market wages and reservation wages. Figure 6 shows the predicted log wage from our model for participants and non-participants and the actual log wage for participants through time. We are able to reproduce the time paths of the wages between 1980 and 2000 for the participants. For the non-participants, the predicted log market wage follows that of participants closely, at a slightly lower level. 14 In our analysis below, when we construct the shadow price of ingredients, we use the predicted wage. For non-participants, we use the maximum of the predicted wage and predicted reservation wage, as their value of time must be greater than their potential wage if they choose not to work.

Shadow price of home-cooked food
We use the estimates from the wage equation to calculate the shadow price of home-cooked food to see how this has change over the period. The shadow price of home-cooked food is given by .

F I G U R E 7
Prices of ingredients and processed food, and shadow price of home-cooked food Note: All prices are relative to the price of the outside good, and assume the outside good is private.
We calibrate the returns-to-scale parameter = 0.8 and B = 0.88 so that the average observed expenditure on ingredients almost exactly matches the average reported time on food management, separately for households without children and those with two children. 15 When computing the shadow price of home-cooked food, we use the observed wage for the participants and the maximum of the market wage and reservation wage for non-participants. We show their time path in Figure 7, together with the market prices of ingredients and of processed foods. We express all these prices relative to the price of an outside good. The outside good includes all non-durable non-food consumption, including alcohol, tobacco, household services, personal goods and services, and leisure goods and services. For participants on average, the cost of time constituted 57 per cent of the shadow price in 1980, rising to 73 per cent in 2000. For non-participants, the share of time cost in shadow price increased from 62 per cent to 75 per cent over the same period. Wages have grown whilst the prices of foods have decreased, and because cooking takes time, the shadow price of home-cooked food, which incorporates the opportunity cost of time, has in fact increased over the period, as is shown in the dashed lines in Figure 7. The increase is 14 per cent for labour market participants and 5 per cent for non-participants. Thus, overall, the true cost of homecooked food has increased in the UK over the sample period.
There are a number of caveats we should make to this result. Our estimate of the shadow price of home-cooked food could be mismeasured for a number of reasons. In our view, it is unlikely that these are large enough to reverse our claim that the shadow price has increased over time, but we are not able to show this conclusively in this paper.
One issue is that fertility choices are endogenous and likely correlated with female labour supply decisions. This introduces a negative correlation between female work and economies of scale in cooking, and may bias our estimates of the shadow cost of home-cooking, due to differential wages between part-time and full-time work. This is a limitation of our analysis that could potentially be addressed by modelling differential female wages in part-time and full-time work. We leave that analysis for future work.
A second issue is that it is generally difficult to predict the cost of time for non-working women. Conceptually, the marginal cost of time for an individual who has chosen not to work might even be lower than their market wage, due to fixed costs of working. In addition, there could be unobserved cross-sectional differences in preferences, which means the finding that working women cook less is not necessarily inconsistent with them having a higher or lower cost of time.
The fact that we find an increasing shadow price for working women and declining consumption of ingredients among them provides some support for our interpretation. In addition, our imputed wages for non-working women are likely to overestimate their marginal cost of time. We use the maximum of the market wage and reservation wage for non-participants. Due to fixed costs of working, once they have decided not to work, their marginal cost of time might be lower than both their market wage and their reservation wage. The proportion of non-working women decreases over time, so the extent of overestimation of the mean shadow price for this reason would decrease over time, and so the true mean shadow price would increase more than in our calculation.
These are interesting issues that merit further investigation, but are beyond the scope of this paper. While there may be systematic errors in the imputed time cost for non-participants, we believe that it is unlikely to affect our main conclusion in this paper.

SUMMARY AND FINAL COMMENTS
There has been a significant decline in UK households' consumption of ingredients in the past 30 years, both in terms of real quantities and relative to the total food consumption. This has occurred despite a long-term fall in the price of ingredients relative to processed food. The key to understanding this phenomenon is to recognise that the true cost of ingredients includes the opportunity cost of cooking time, which has increased rapidly due to wage growth. We have written down a simple model of consumption that explicitly incorporates the time cost involved in the home production of food; this allows us to derive the shadow price and to show that this has risen. Thus, our analysis offers an intuitive and data-based explanation to the real decline in consumption of ingredients, without resorting to unobserved shifts in preferences or quality. These facts have implications for policy that aims to encourage healthier diets. Our analysis shows that the shadow price of home-cooked food depends as much on the wage as on the market price. Taxes or subsidies to market prices would have to be large to provide incentives for households to switch away from ready-to-eat food all together and to consume more home-cooked food, unless they were accompanied by changes in time use. These would not have to be a reduction in labour market participation by secondary earners, but could instead be driven by changes in preferences for leisure activities (for example, substituting cooking for other leisure activities).
These results also suggest that policies that lead to improvements in the nutritional characteristics of processed foods, for example, by encouraging firms to reformulate, might be more effective than policies that focus only on increasing home cooking. In terms of future avenues of work, there are many unanswered questions on the relationship between time use and consumption that could fruitfully be explored.

A C K N O W L E D G E M E N T S
We gratefully acknowledge financial support from the European Research Council (ERC) under ERC-2009-AdG-249529 and ERC-2015-AdG-694822, and from the Economic and Social Research Council (ESRC) under the Centre for the Microeconomic Analysis of Public Policy (CPP), grant number RES-544-28-0001. We would like to thank Tom Crossley, who was a coauthor on an earlier version of this work and who helped us all very much, and Zoe Oldfield for her assistance with setting up the data.